MHC相关论文

Balancing and Directional Selection at Exon-2 of the MHC DQB1 Locus among Populations of Odontocete Cetaceans Demetrios Vassilakos,*1 Ada Natoli,* Marilyn Dahlheim,� and A. Rus Hoelzel* *School of Biological and Biomedical Sciences, University of Durham, United Kingdom; and �National Marine Mammal Laboratory, National Marine Fisheries Service, Seattle, WA The diversity of exon-2 (peptide-binding region) of the DQB1 locus (Class II, major histocompatibility complex, MHC) was investigated on an extended sample of populations of three focal cetacean species (two sibling delphinid species and another in the same family). We tested the hypothesis that dolphin populations with a worldwide distribution across different habitats and geographic regions will be under differential selective pressure by comparing DQB1 variation with variation at neutral markers and by investigating putative functional residues within the exon-2 sequence at the population level. Variation at the DQB1 locus was not correlated to neutral differentiation (assessed by comparison with microsatellite DNA markers), and overall FST values were significantly lower for the MHC locus, consistent with expectations for balancing selection. Measures of heterozygosity and dn/ds ratios were also consistent with balancing selection. However, outliers in the FST comparisons and the analysis of putative functional residues suggested incidences of directional selection in local populations. Introduction Themajor histocompatibility complex (MHC) is amul- tigene family that codes for cell surface glycoproteins that bind peptides of processed foreign antigens andpresent them to T-lymphocytes. MHC class I and class II loci have been shown tobe highly polymorphic in, for example, primate, ro- dent, pinniped, avian, andbovine species (Klein 1986;Udina et al. 1994; Trowsdale 1995; Ellegren et al. 1996; Nasir et al. 1997; Wagner et al. 1998; Chardon et al. 1999; Horin and Matiasovic 2002; Otting et al. 2002; Villegas-Castagnasso et al. 2003; Piertney and Oliver 2006). Furthermore, poly- morphism in MHC genes is often greatest at the sites that specify the amino acids of the peptide binding region (PBR), the region that is responsible for peptide collection andpresentation (Klein andFigueroa1986). Twoof themain reasons that MHC polymorphism has been attributed to fre- quency and/or overdominant selection are: 1) the high non- synonymous (dn) relative to synonymous (ds) substitution rates in the PBR and 2) trans-species polymorphism (Nei and Rooney 2005). Neutral theory predicts that genes under selection can behave as effectively neutral when populations are small (when s , 1/2Ne), and therefore, MHC alleles may experi- ence periods of neutral evolution, during which genetic drift and mutation are more prominent in maintaining MHC polymorphism thanselection.MHCpopulationgenetic stud- ies on Scandinavian beavers (Ellegren et al. 1993), bighorn sheep (Boyce et al. 1997), Amerindians (Cerna et al. 1993; Valdes et al. 1999), Australian bush rats (Seddon and Baverstock 1999), Atlantic salmon (Bernatchez and Landry 2003), wolves (Canus lupus; Seddon and Ellegren 2004), mountain goats (Oreamnos americanus; Mainguy et al. 2007), and brown trout (Salmo trutta; Campos et al. 2006) among others show evidence for the influence of genetic drift, often together with evidence for balancing selection. Landry and Bernatcez (2001) showed that in the At- lantic salmon, differences in the environment and geo- graphical scales correlated with significant differences in MHC class II B allelic frequencies. However, overall FST values were not significantly different from values for microsatellite loci. They suggested that, although bal- ancing selection was evident, the population structure in- ferred by MHC analysis was shaped more by genetic drift and migration than selection (Landry et al. 2001). Campos et al. (2006) reported similar results for brown trout. Seddon and Ellegren (2004) found significant differ- entiation at DQA, DQB1, and DRB1 loci (FST 5 0.251– 0.269) among wolves in similar habitats in Finland, Estonia, Latvia, and eastern Russia. However, they also showed that temporal changes in variability at these loci were for the most part consistent with neutral evolution. The authors suggest that this was due to fragmentation and consequent genetic drift (Seddon and Ellegren 2004). Hayashi et al. (2006) also found evidence for both balancing selection overall, and ge- netic drift in small, local populations for the DQB locus in the finless porpoise (Neophocaena phocaenoides). Evidence for directional selection has come in part from the mapping of allelic substitutions onto the inferred structural model of the MHC molecule (Hughes et al. 1996; Ou et al. 1998; Cohen 2002). For example, a study on the effects of pollution on MHC variation in estuarine fish showed that the population that had adapted to severe chemical pollution had specific amino acid substitutions in the a-helix region (Cohen 2002). Furthermore, the fish from the unpolluted area also exhibited a significantly dif- ferent pattern in the b-pleated sheet of the PBR (Cohen 2002). Functional analysis has also been used in human studies (with the potential to extend this work to nonhuman species), where human pathologies have been correlated to specific amino acid replacement and motif changes in the PBR among different populations (e.g., Nepom and Erlich 1991; Winchester 1994; Hill 1998; Ou et al. 1998). Evi- dence for positive selection based on geographic or tempo- ral differences in allele frequencies has been reported for transporter associated with antigen processing (TAP) genes in brown trout (Jensen et al. 2008) and for class II MHC genes among populations of the Asian cygomolgus ma- caque (Macaca fascicularis; Bonhomme et al. 2007), among other studies. TAP proteins deliver cytosolic 1 Present address: Division of Molecular Immunology, MRC National Institute for Medical Research, London, United Kingdom. Key words: evolution, population genetics, marine mammal, MHC, immune system. E-mail: a.r.hoelzel@dur.ac.uk. Mol. Biol. Evol. 26(3):681–689. 2009 doi:10.1093/molbev/msn296 Advance Access publication January 6, 2009 � The Author 2009. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org peptides to the endoplasmic reticulum where they associate with the MHC molecule for presentation and have been shown to coevolve with MHC Class I molecules (McCluskey et al. 2004). Pathogens that affect marine mammals have been as- sociated with the marine environment for long evolutionary periods (Howard et al. 1983; Kennedy 1990; Limpscomb et al. 1994; Higgins 2000). For example, in cetaceans, the divergence between different morbillivirus species and a hypothetical common terrestrial ancestor occurred millions of years ago (Barrett et al. 1993, 1995; Osterhaus et al. 1995). Cetaceans are warm blooded and breathe air like all mammals; however, they rely on the aquatic envi- ronment for their life needs. This may result in an interac- tion with both terrestrial and marine pathogen risks. In the last 20 years, thousands of marine mammals have died due to epizootics caused by viral infections (Van Bressem et al. 1999). In many cases, these are likely indigenous patho- gens, though not all. The Canine Distemper Virus infection from dogs, which took place in 1986 at lake Baikal in Rus- sia, had devastating effects on the freshwater Baikal seal population (Mamaev et al. 1996; Forsyth et al. 1998). Killer whale (Orcinus orca) populations can be found across all major oceans in both polar and temperate waters and in particular in coastal areas of high productivity (Dahleim and Heyning 1999; Ford 2002; Hoelzel, Natoli, et al. 2002). The social associations formed by this species are very stable, and there are regional populations that are known to have persisted for decades (Ford et al. 1998; Ford 2002; Hoelzel, Natoli, et al. 2002). Sympatric populations of foraging specialists (different ecotypes pursuing fish vs. marine mammal prey) found in the eastern North Pacific differ in ecology, behavior, and distribution patterns and are genetically differentiated, as are populations of the same ecotype in parapatry and allopatry (Hoelzel and Dover 1991; Hoelzel, Dahleim, and Stern 1998; Ford and Ellis 1999; Hoelzel, Goldsworthy, and Fleischer 2002). Popula- tions of the fish-eating ecotype have been referred to as ‘‘residents’’ and the marine mammal–eating ecotype as ‘‘transients,’’ and this terminology will be used here. Ge- netic analyses have identified at least seven populations in the North Pacific (Hoelzel et al. 2007). The bottlenose dolphin (Tursiops truncatus) is also found in all major oceans, from cold temperate to tropical seas, in coastal and offshore waters. Tursiops truncatus ex- hibits habitat differentiation among populations across its range, as sympatric or parapatric populations will use the coastal (nearshore) or the pelagic (offshore) environment (Hoelzel, Potter, and Best 1998; Hoelzel, Goldsworthy, and Fleischer 2002; Natoli et al. 2004). Studies on mtDNA and nuclear DNA (microsatellites) have shown that coastal and pelagicT. truncatus populations in thewesternNorthAtlantic are genetically differentiated (Hoelzel, Dahleim, et al. 1998). In addition, inSouth Indian andSouthPacific coastal habitats, a smaller morphotype has been described as the ‘‘aduncus’’ form. An mtDNA study byWang et al. (1999) demonstrated that the coastal aduncus form in China shows a reciprocally monophyletic relationship to the offshore populations of T. truncatus, supporting the classification of the aduncus form as a separate species, Tursiops aduncus. Further to this, Natoli et al. (2004) showed that the South African aduncus morphotype formed a monophyletic lineage separate from both T. truncatus and the Chinese aduncus type. Genetic differentiation between coastal and offshore populations (Hoelzel et al. 1998) and among putative populations in the eastern North Atlantic, Mediterranean, and Black Sea sug- gested differentiation determined by oceanic habitat bound- aries (Natoli et al. 2005). Here, we test the hypothesis that for these species (each of which show neutral genetic differentiation appar- ently driven by differences in habitat or foraging strategy) regional populations will show differential evidence of bal- ancing or positive selection at the DQB1 MHC locus. In support of this, we find evidence for MHC differentiation that is not consistent with isolation by distance models or differentiation patterns seen at presumably neutral microsa- tellite DNA loci. We also chose a specific set of residues within the PBR known to show adaptive differentiation in other species, and found patterns consistent with differ- ential selection for some regional populations. The impli- cation is that both balancing and local positive selection are important in determining the pattern of variation at this locus in these species. Materials and Methods Samples Tissue samples were acquired from various sources and extracted to DNA by standard methods (see Natoli et al. 2004, 2005; Hoelzel et al. 2007 for details). Killer whale samples were from the ‘‘southern resident’’ popula- tion off Washington state (SR; N 5 33; see Hoelzel et al. 2007), Alaskan residents off SE Alaska (AR; N 5 31), Alaskan transients (AT; N 5 35), Californian transients (CT; N 5 24), the Bering Sea and Aleutians (BR; N 5 14), and Iceland (IC; N 5 31). Tursiops truncatus popula- tions were from the Mediterranean Sea (MED; N5 29; see Natoli et al. 2004), Eastern North Atlantic (ENA; N 5 26), Western North Atlantic pelagic (WNAP; N5 25), Western North Atlantic coastal (WNAC; N 5 27), and the eastern North Pacific off southern California (ENP; N 5 15). A T. aduncus population was sampled off South Africa (SAA; N 5 140). Map locations are provided in figure 1. Molecular Methods The exon-2 PBR region was amplified using the pri- mers (CTGGTAGTTGTGTCTGCACAC and CATGTGC- TACTTCACCAACGG) developed by Tsuji et al. (1992). The reaction conditions were 10 mM Tris HCl (pH 5 8.3), 50 mM KCl, 2.5 mM MgCl2, 0.2 mM of each dNTP, 0.25 lM of each primer, 2 units of Pfu Taq polymerase (Promega, Southampton, UK), and 100–150 ng of template DNA in a 25-ll final volume. For screening by SSCP, 2 ll of denaturing loading buffer 95% (v/v) formamide, 0.1% (w/v) bromophenol blue, 0.1% (w/v) xylene cyanol, and 10 mM NaOH (Sigma–Aldrich, Gillingham, UK) were added to 2 ll of PCR product and were loaded on a nonde- naturing acrylamide gel 6% (v/v) 49:1 acrylamide:bis- acrylamide, 10% (v/v) glycerol, and 1� TBE for 6 h and 40 W migrations at 4 �C. The gel was incubated for 682 Vassilakos et al. 20 min with the fluorescent GelStar Nucleic Acid Gel Stain (BioWhittaker, Rockland, ME) according to manufacturer instructions. Allelic conformation was visualized by expo- sure to short-wave UV light and photographed. To confirm apparent genotypes and test for possible Escherichia coli recombinant artifacts (Longeri et al. 2002), up to 20 clones were rescreened by SSCP from different individuals and a subset sequenced in both directions (including the se- quence of every allele for multiple individuals). Cloning was done using the Easy T-Vector Cloning kit (Promega) according to the manufacturer instructions. The polymerase chain reaction (PCR) fragments were purified using a PCR purification kit (Qiagen, Crawley, UK) and inserted into the EcoR1 site of the pGEM-T vector plasmid. Because the PCR fragments were generated using Pfu DNA polymerase (Promega) they were blunt ended, and therefore an A-Tail reaction was required. Sequencing was performed by the Big-Dye terminator reaction using the universal sequencing primers of the Easy T-Vector plasmid. Structural Analysis Human leukocyte antigen (HLA) DR (and DQ; Wecherpfennig and Strominger 1995) has identified subre- gions (referred to as pockets) in the binding groove, which influence binding, presentation, and recognition by T-cell receptors (Stern et al. 1994). Among these subregions, pocket P4 amino acid residues b70, b71, and b74 have been shown to play a significant role in determining T-cell rec- ognition of the peptide–HLA complex (Olson et al. 1994; Stern et al. 1994; Ou et al. 1996). It has been shown through site-directed mutagenesis inDQ andDR alleles of HLA that selective peptide binding is greatly affected by the amino acid residues in pocket P4 and the consequent charge (Hammer et al. 1995; Wecherpfennig and Strominger 1995; Ou et al. 1996). Alleles have been grouped into seven different functional categories according to physicochemi- cal polymorphisms of these residues, and their influences on T-cell receptor recognition (Ou et al. 1997, 1998). Ou et al. (1998) suggested that these seven categories can be combined into four groups based on the sum of the charges at the b70, b71, and b74 residues: a positively charged group (þ), a negatively charged group (�), a dicharged group (þ/�), and a neutral group (n). For example, the DRB1*1117 allele exhibits the residues RRE in positions 70, 71, and 74, respectively. Arginine (R) is positively charged, and glutamic acid (E) is negatively charged and so this allele is classified in the dicharged functional group. When a charged amino acid is present among nonpolar and/ or neutral amino acids, then the allele is classified according to the charged amino acid present. The charge of the amino acids was determined according to the following categori- zation (Ou et al. 1998): H, K, and R positive; D and E neg- ative; and the rest neutral. Comparisons of charge profiles were done using contingency tables implemented in the program RxC (http://www.marksgeneticsoftware.net/; 20 batches, 2,500 replicates per batch). RxC employs the metropolis algorithm to obtain an unbiased estimate of the exact P value (see Raymont and Rousset 1995). Population Genetic Analysis The allele frequencies, allelic richness, and gene diver- sity index (Hs) of the DQB1 locus for each of the popula- tions were estimated using FSTAT version 2.9.3 (Goudet 2001). The expected allelic frequencies under neutrality were estimated by the Ewens–Watterson–Slatkin exact test using the program ARLEQUIN version 2.000 (Schneider et al. 2000). In addition, ARLEQUIN was used to perform the Mantel matrix correlation test, FST index, expected (He based on the Hardy–Weinberg equilibrium) and observed (Ho) genotype frequencies. Statistical significance was es- timated by a Chi-square test (P , 0.05, after Bonferroni correction). The Nei–Gojobori method (implemented in MEGA) was used to estimate the dn/ds ratio within the PBR region of the DQB1 sequence. FST values for the DQB1 locus were compared with published data on 16 mi- crosatellite DNA loci for the killer whale (Hoelzel et al. 2007) and nine microsatellite DNA loci for the bottlenose dolphin (Natoli et al. 2004, 2005). DISTLM v.5 (Anderson 2004) was used to perform a permutation test for the FST matrices corrected for ln transformed geographic distance (Anderson 2001). Bootstrapping (15,000 replications) was undertaken over all microsatellite DNA loci to calcu- late the 95% confidence intervals (CIs) around FST esti- mates (using FSTAT). MHC FST values outside the 95% CIs were considered significantly different from the estimates derived using microsatellites (after Landry and Bernatcez 2001). FST values are correlated with heterozygosity levels, so that outliers from this relationship can suggest direc- tional (FST higher than expected) or balancing selection (FST lower than expected; Beaumont and Nichols 1996). We tested this using FDIST (Beaumont and Nichols 1996) as implemented through LOSITAN (Antao et al. 2008). Simulations were run for 10,000 replications, FIG. 1.—Sample locations for Tursiops sp. (in gray): 1: SAA, 2: ENA, 3: MED, 4: ENP, 5: WNAC, and 6: WNAP; for killer whales (in black): 1: SR, 2: AR, 3: AT, 4: CT, 5: IC, and 6: BR (see text for definition of abbreviations). Population Genetics of Odontocete MHC DQB 683 95% CI, and using the options for neutral and forced mean FST. Outlier microsatellite loci were omitted (one for each species), as suggested by the authors, though this made no difference to the position of the MHC locus relative to the confidence limits in either case (data not shown). An infinite allele model was assumed, but replications using the step- wise mutation model made no difference to the result (data not shown). Bottlenose dolphin comparisons using this test were for T. truncatus only. Results Tables 1 and 2 summarize the data on indicators of DQB1 diversity and possible selection in each population for thekillerwhale (table1) andbottlenosedolphins (table2). All killer whale populations showed a significant excess of observed heterozygotes compared with Hardy–Weinberg expectations (with all FIS values significantly negative), and the level of diversity was similar among populations. For the bottlenose dolphin, four out of six populations showed a significant deficit of heterozygotes, whereas 2 showed significant excess. Evidence for balancing selection based on the Ewens–Watterson neutrality test has low power, but three populations were significant at the P , 0.05 level for the killer whale and one for the bottlenose dolphin. All populations of both species showed a PBR dn/ds ratio that was significantly greater than 1, with the bottlenose dolphin populations showing the strongest effect (tables 1 and 2). The bottlenose dolphin sample is rep- resented by two sibling species, T. truncatus (represented by five populations) and T. aduncus (represented by one popu- lation). The latter is well sampled (N 5 140) and provides clear evidence for heterozygote excess, whereas the smaller T. truncatus sa