Spatial variation and low diversity in the major histocompatibility complex in walrus (Odobenus rosmarus)
- First Online:
- Cite this article as:
- Sonsthagen, S.A., Fales, K., Jay, C.V. et al. Polar Biol (2014) 37: 497. doi:10.1007/s00300-014-1450-9
- 247 Views
Increased global temperature and associated changes to Arctic habitats will likely result in the northward advance of species, including an influx of pathogens novel to the Arctic. How species respond to these immunological challenges will depend in part on the adaptive potential of their immune response system. We compared levels of genetic diversity at a gene associated with adaptive immune response [Class II major histocompatibility complex (MHC), DQB exon 2] between populations of walrus (Odobenus rosmarus), a sea ice-dependent Arctic species. Walrus was represented by only five MHC DQB alleles, with frequency differences observed between Pacific and Atlantic populations. MHC DQB alleles appear to be under balancing selection, and most (80 %; n = 4/5) of the alleles were observed in walruses from both oceans, suggesting broad scale differences in the frequency of exposure and diversity of pathogens may be influencing levels of heterozygosity at DQB in walruses. Limited genetic diversity at MHC, however, suggests that walrus may have a reduced capacity to respond to novel immunological challenges associated with shifts in ecological communities and environmental stressors predicted for changing climates. This is particularly pertinent for walrus, since reductions in summer sea ice may facilitate both northward expansion of marine species and associated pathogens from more temperate regions, and exchange of marine mammals and associated pathogens through the recently opened Northwest Passage between the Atlantic and Pacific Oceans in the Canadian high Arctic.
KeywordsGenetic diversityMajor histocompatibility complexOdobenus rosmarusPopulation genetic structureWalrus
The impact of climate change has been greatest at high latitudes. The rate of Arctic warming has been two to three times greater than the global average over the past 150 years (Trenberth et al. 2007), and this trend is expected to continue (Prowse et al. 2009). Changes in the Arctic over the past three decades are consistent with predictions associated with global climate warming, including a 45,000 km2/year reduction in seasonal minimal ice extent (Moore 2006; Parkinson and Cavalieri 2008), and earlier seasonal ice breakup and later freezing (Stirling and Parkinson 2006). Future predictions indicate the possibility of nearly sea ice-free summers in the Arctic Ocean within 20 years (Overland and Wang 2013). Increased temperature coupled with changes in habitat will likely result in a northward advance of species, altering ecosystem dynamics, and species interactions (Post et al. 2009), including an influx of pathogens novel to the Arctic and a potential increase in virulence (Kutz et al. 2005; Jenkins et al. 2006; Hoberg and Brooks 2008). Because parasite and pathogen diversity are generally lower in high latitudes, species living there are likely to be immunologically naïve to pathogens accompanying species that may expand from temperate regions. For high latitude marine species, the opening of the Northwest Passage in the Canadian high Arctic may also increase opportunities for pathogen transfer associated with dispersal between the Pacific and Atlantic Oceans.
The response of individuals to immune challenges is mediated by both the innate and adaptive immune systems. The innate immune system, including epithelial barriers, phagocytic and natural killer leukocytes, and dendritic cells, is always present and ready to mobilize. In contrast, the adaptive immune system, including certain peptides encoded by the major histocompatibility complex (MHC), is a silent component of the immune system that is activated when the innate immune defenses are evaded or overrun. Both the innate and adaptive immune systems are generally considered equally important aspects of an individual’s immune response. However, because populations in the Arctic are typified by a naïve adaptive immune system (Hoberg and Brooks 2008), the near-term response of Arctic species is likely to be more greatly dependent on the ‘adaptive potential’ of their immune response system.
The ability of individuals (and populations) to respond to disease and pathogens is thought to be associated with genetic diversity at the MHC, a family of genes involved in adaptive immune response in vertebrates (see Sommer 2005 for review). High MHC variation observed within populations is often attributed to exposure to a wide range of disease challenges coupled with the active maintenance of multiple alleles (i.e., balancing selection; Edwards and Potts 1996). Therefore, populations with greater diversity at MHC genes are hypothesized to have increased ability to respond to novel pathogens (Doherty and Zinkernagel 1975; Hughes and Nei 1988). Specifically, divergent allele combinations are likely more advantageous for carriers as they can present disparate antigens during adaptive immune response (i.e., divergent allele advantage; Wakeland et al. 1990). This mechanism may explain not only the high number of MHC alleles within populations but also the presence of divergent alleles within the individual (Lenz et al. 2009). Maintenance of high levels of MHC allelic diversity is likely not optimal within species or populations that are only exposed to a single pathogen (Wedekind et al. 2005). Therefore, reduced MHC diversity within species is often attributed to low pathogen exposure over time, such as in the Arctic, or a recent severe disease mortality event (Slade 1992). However, low MHC diversity does not always equate to increased disease susceptibility. Some species, such as those with wide-ranging distributions and/or migratory behavior that would presumably expose individuals to a broad spectrum of disease diversity, nevertheless exhibit reduced MHC diversity that may be associated with high efficiency of innate immune response (e.g., Falco peregrinus; Gangoso et al. 2012).
Arctic species will likely face a broad range of ecological and immunological challenges as the global climate warms. Sea ice-dependent species, such as the walrus (Odobenus rosmarus), will likely be exposed to some of the most rapid ecological changes due to global climate change associated with changes in seasonal sea ice dynamics and nutrient cycling (Post et al. 2009). The walrus is discontinuously distributed around the Arctic Basin, and within each occupied region, a subspecies has been described (Atlantic, O. r. rosmarus; Laptev Sea, O. r. laptevi; and Pacific, O. r. divergens; Chapskii 1940; Fay 1985), though O. r. laptevi may represent the westernmost population of O. r. divergens (Lindqvist et al. 2009). Walruses are highly vagile and most populations migrate seasonally with the advance and retreat of sea ice. The walrus is a gregarious species that forms concentrated breeding areas and non-breeding aggregations. This gregarious lifestyle lends itself to possible increased exposure to high levels of pathogens due to the close proximity of individuals and accumulation of fecal material (Cammen et al. 2011). Approximately 90 % of walruses occur in the North Pacific (Fay 1985), with breeding activities concentrated in three main areas in the Bering Sea, and located fewer than 800 km apart. By contrast, within the North Atlantic Ocean, walruses are distributed across a larger region, with some populations up to 2,800 km apart (Born et al. 2001). Further, walrus habitat use varies between the Atlantic and Pacific Oceans, with Atlantic walruses favoring coastal areas and Pacific walruses, notably females and calves, utilizing ice floes (Laidre et al. 2008). Genetic studies suggest that variances in allelic and haplotypic frequencies between breeding populations of Pacific walruses are tenfold lower (Sonsthagen et al. 2012) than between Atlantic walrus populations (Born et al. 2001). Examination of genes associated with immune response may help elucidate whether there have been differential evolutionarily relevant and adaptive processes operating within and between Atlantic and Pacific walruses.
The combination of differences in the spatial distribution of breeding areas, concentration of walruses across their range, habitat use, and levels of genetic structure in walruses of the northern Pacific Ocean relative to the Atlantic Ocean suggests that walruses in the two regions may also have differing levels of MHC diversity, and therefore differences in their ability to respond to diverse immune challenges. Here, we compared levels of genetic diversity at Class II MHC locus DQB exon 2 among walrus populations relative to a baseline level of genetic diversity at neutral markers.
Genotyping and sequencing
Genotype data from the Atlantic walrus were obtained for the same 11 microsatellite loci as Sonsthagen et al. (2012; Orr2, Orr3, Orr7, Orr8, Orr11, Orr16, Orr21, Orr24, Buchanan et al. 1998; Hg6.1, Allen et al. 1995; 1GF-1, Kirpatrick 1992; and SgPv9, Goodman 1997). PCR amplifications were carried out in five multiplex reactions using procedures similar to those described in Cronin et al. (2009). Electrophoresis of PCR products, gel standardization, and sizing of microsatellite alleles follow Sonsthagen et al. (2004). Ten percent of the samples were genotyped in duplicate for the 11 microsatellite loci for quality control.
For the Atlantic walrus populations, we amplified a 589–592 base pair (bp) segment of mtDNA comprising 106 bp of the cytochrome b gene, 70 bp of the tRNA-thr, 65 bp of the tRNA-pro, and 348–351 bp of the hypervariable portion of the control region, using primers described elsewhere (L15774b and H00019; Sonsthagen al. 2012). Data (microsatellite genotypes and mtDNA sequences) for the Pacific walrus populations were collected by Sonsthagen et al. (2012).
MHC Class II genes are involved in adaptive immune response to extracellular parasites and pathogens, and exon 2 is the most polymorphic in the DQ MHC genes (Meyer-Lucht and Sommer 2005), making it a good candidate for assessment of genetic diversity at the population level as well as a predictor of adaptive immune potential (Acevedo-Whitehouse and Cunningham 2006). Exon 2 of the MHC Class II DQB locus was amplified with primer pairs DQBF (5′-GATTTCGTGTACCAGTTTAAGGGC-3′) and DQBR (5′-CCACCTCGTAGTTGTGTCTGCA-3′) for the Atlantic and Pacific populations. PCR amplifications, cycle-sequencing protocols, and post-sequencing processing followed Sonsthagen et al. (2004), except DQB sequences that contained double peaks of approximately equal peak height, indicating the presence of two alleles, were coded with International Union of Pure and Applied Chemistry (IUPAC) degeneracy codes and treated as polymorphisms. Heterozygotes were inferred from the sequence data but confirmed using single-stranded conformation polymorphism (SSCP; Sunnucks et al. 2000), following methods outlined in Gangoso et al. (2012). Confirmation and resolution of the gametic phase of DQB alleles were straightforward because the alleles (except DQB*05) were represented by homozygous individuals. For quality control purposes, we extracted, amplified, and sequenced 10 % of the samples in duplicate. No inconsistencies in nucleotide base calling or SSCP allele configurations were observed between replicates. Sequences were accessioned in GenBank (mtDNA KJ004382–KJ004393; MHC KJ004394–KJ004398).
Estimation of genetic diversity
Allelic richness, observed and expected heterozygosities, Hardy–Weinberg equilibrium (HWE), and linkage disequilibrium were calculated for each microsatellite locus in GENEPOP 3.1 (Raymond and Rousset 1995) and FSTAT version 2.9.3 (Goudet 1995). Haplotype (h) and nucleotide (π) diversity for mtDNA and DQB sequence data were estimated in ARLEQUIN 2.0 (Schneider et al. 2000). We performed tests of selective neutrality and historical fluctuations in population demography in ARLEQUIN using Fu’s FS (Fu 1997) and Tajima’s D (Tajima 1989) at mtDNA. Both Tajima’s D and Fu’s FS are sensitive to historical fluctuations in population size; however, Fu’s FS critical significance values of 5 % require a P value below 0.02 (Fu 1997). An unrooted phylogenetic network for mtDNA control region and DQB sequence data were constructed from haplotypes and alleles in NETWORK 4.610 (Fluxus Technology Ltd. 2009) using the reduced median method (Bandelt et al. 1995), to illustrate possible reticulations in the gene tree because of homoplasy or recombination.
We used the program MEGA version 5.05 (Tamura et al. 2011) to assess the relative rates of synonymous (dS) to non-synonymous (dN) substitutions at nucleotide positions at DQB according to Nei and Gojobori (1986) using the Jukes–Cantor (1969) correction. These rates were calculated for the entire coding regions, as well as for ABS and non-ABS regions separately and tested for significance using the Z test, as implemented in MEGA. We further tested for evidence of positive selection using the maximum likelihood methods implemented in CODEML in program of PAML version 4.2 (Yang 1997, 2007) by fitting two models of codon substitution: M1 (nearly neutral) and M2 (selection). Bayesian posterior probabilities were calculated for positively selected sites using Bayes empirical Bayes (BEB) for model M2. Log-likelihood ratio tests (LRT) were performed to compare the two models with and without selection (M2 versus M1). Statistical significance is determined by comparing twice the log-likelihood scores (2ΔLnL) to a χ2 distribution with degrees of freedom equal to the difference in the number of parameters between the models (Yang 1997).
Estimation of spatial genetic structure
Spatial variance in allelic and haplotypic frequencies was calculated (FST and ΦST, respectively) in ARLEQUIN, adjusting for multiple comparisons using Bonferroni correction (α = 0.05). Because the upper possible FST value for a set of microsatellite loci is usually < 1.0 (Hedrick 2005), we used RECODEDATA version 1.0 (Meirmans 2006) to calculate the uppermost limit of FST for a given data set. Pairwise ΦST was calculated using a Hasegawa, Kishino, Yano nucleotide substitution model (HKY; Hasegawa et al. 1985) as determined using MODELTEST (Posada and Crandall 1998), and Akaike’s information criterion (Akaike 1974). Hierarchical analyses of molecular variance (AMOVA) were conducted between Pacific walrus and Atlantic walrus populations to assess levels of differentiation between walrus subspecies in ARLEQUIN.
Genetic diversity indices based on genotype information for 11 microsatellite loci and sequence information from mitochondrial DNA control region and major histocompatibility complex locus DQB assayed from five populations of walruses, including; allelic richness (AR), number of alleles (A), observed and expected heterozygosity (H0/He), number of mtDNA haplotypes (H), haplotype (h) and nucleotide (π) diversity, and tests of selective neutrality (Tajima’s D and Fu’s FS)
St. Lawrence (n = 32)
SE Bering (n = 34)
Thule (n = 10)
Attu–Sisimiut (n = 10)
Scoresby Sound (n = 10)
The relative frequency of non-synonymous substitutions (dN = 0.215, SE = 0.088) was larger than the frequency of synonymous substitutions (dS = 0.000, SE = 0.000) at the ABS for all alleles, with a dN/dS ratio greater than one (P = 0.004, Z = 2.945) suggestive of selection for diversity (balancing selection) at these positions (Tamura and Nei 1989). Non-synonymous substitution occurred less frequently than synonymous substitutions, resulting in a dN/dS ratio less than unity, in the non-ABS region (dN = 0.020, SE = 0.012; dS = 0.064, SE = 0.028; P = 0.084, Z = 1.740), as well as across the entire sequence (dN = 0.049, SE = 0.016; dS = 0.053, SE = 0.024; P = 0.882, Z = 0.149). The site-specific analyses in PAML identified positively selected sites in DQB and LRT revealed that the selection model fit the data better than the neutral model (M1 LnL = −394.5, M2 LnL = −385.9; χ2 = 17.2, df = 2, P < 0.005). The Bayesian approach in model M2 revealed two positively selected sites at positions 13 (P = 0.995, where P denotes posterior probability of assigning a site to the positively selected class) and 44 (P = 0.991). Amino acid site 44 corresponds to an ABS (Table 2).
Spatial genetic structure
Genetic structure (FST, RST, and ΦST) among walrus populations at autosomal microsatellite loci, mitochondrial DNA (mtDNA) control region and Class II DQB major histocompatibility complex (MHC) locus across the entire sequence as well as only among sites that are antigen-binding sites (ABS)
Genetic structure observed at MHC between Pacific walrus and Atlantic walrus populations was intermediate between the two neutral marker types; FST ranged from 0.322 to 0.666 and ΦST ranged from 0.150 to 0.451 (Table 3). Similarly, AMOVA results were intermediate relative to the neutral markers (FST = 0.469, P < 0.001; ΦST = 0.294, P < 0.001). Levels of genetic differentiation were lower when based on variation only at MHC ABS for both interpopulation comparisons (FST ranged from −0.006 to 0.465, and ΦST ranged from −0.006 to 0.274; Table 3) and the AMOVA grouping (FST = 0.195, P < 0.001; ΦST = 0.138, P < 0.001).
After accounting for the maximum possible FST for our suite of microsatellite loci (Hedrick 2005) and differences in the effective size between nuclear and mitochondrial genomes [FST(nu) = 1 − e0.25*ln[1−FST(mt)]; Zink and Barrowclough 2008], levels of genetic structure between Pacific and Atlantic walrus populations at MHC alleles were higher than neutral expectation for half of the population comparisons (n = 20/40, 50 %; Fig. 3). However, only two comparisons (St Lawrence and Attu–Sisimiut; SE Bering and Attu–Sisimiut) between Pacific and Atlantic walrus populations exhibited higher levels of structure based on both MHC alleles and ABS than neutral expectation (Fig. 3). Interestingly within the Atlantic walrus, comparisons between western and northwestern Greenland populations (Thule and Attu–Sisimiut) and eastern Greenland for MHC alleles and ABS exhibit greater levels of structure than expected based on neutral data (Fig. 3).
Only five MHC Class II DQB exon 2 alleles were observed between walruses assayed from Pacific and Atlantic regions. This number of alleles is low relative to terrestrial species (Villanueva-Noriega et al. 2013), but similar to other marine mammals; e.g., between two and eight DQB alleles have been reported in six species of pinnipeds (Hoelzel et al. 1999; Lento et al. 2003; Weber et al. 2004; Cammen et al. 2011). Several mechanisms have been proposed to explain low levels of polymorphism at DQB among marine mammals, including population bottlenecks, reduced selective pressure of pathogens in marine environments, and low rates of mutation (Villanueva-Noriega et al. 2013). Although Pacific walrus underwent a reduction of 50–75 % of individuals between 1880 and 1950 s (Fay 1982; Fay et al. 1997), genetic signatures of population decline were not detected in either microsatellite allelic or mtDNA haplotypic data (Sonsthagen et al. 2012). Therefore, the population decline experienced by walruses is a less likely explanation for the observed reduced DQB polymorphism than low pathogen exposure over time (Slade 1992; Wedekind et al. 2005). Alternatively, low levels of DQB polymorphism in walruses may also suggest that alleles present in these populations may not be pathogen-specific, but rather able to respond broadly to immunological challenges, or a high efficiency of the innate immune response system.
Variance in allelic frequencies at DQB differed between Pacific and Atlantic walruses, and Atlantic walruses residing in western Greenland exhibited greater heterozygosity at DQB than eastern Greenland and Pacific walrus populations. Genetic structure at DQB was greater than neutral expectation among Pacific and Atlantic populations and among western Greenland and eastern Greenland populations (Fig. 3). The Pacific walrus was predominately represented by a single DQB allele and eastern Greenland by two alleles (that do not differ at ABS), with DQB allele 4 absent from eastern Greenland. In contrast, western and northwestern Greenland walruses were represented by four alleles, in approximately equal frequency in western Greenland (Fig. 1). Differences in the frequency of DQB alleles across the species’ distribution indicate that Pacific walruses and eastern Greenland walruses may experience pressure from a smaller spectrum of pathogens with infrequent exposure to other pathogens; this would maintain the presence of other alleles within the population, albeit in low frequency. In contrast, Atlantic walruses in western Greenland may be exposed to similar pathogens though no one pathogen pressure is constant over time. Alternatively, they may experience more frequent exposure across a greater spectrum of pathogen types, which may maintain increased DQB heterozygosity within individuals. Therefore, we hypothesize that broad scale differences in the frequency of exposure and diversity of pathogens are likely influencing levels of heterozygosity at DQB in walruses.
Although the levels of polymorphism observed are low, the DQB alleles appear to be under balancing selection, as indicated by the greater rate of non-synonymous versus synonymous substitutions at ABS (Tamura and Nei 1989), and identification of two sites (one at an ABS) under selection. Villanueva-Noriega et al. (2013) found that cetaceans exhibit weaker balancing selection in DQB relative to terrestrial mammals (artiodactylids and primates) examined. However, the strength of selection varies within cetaceans, suggesting that ecological contexts (coastal vs. pelagic and warm vs. cold environments) may influence the strength of selection at DQB (Villanueva-Noriega et al. 2013). Despite differences in habitat use within walruses, four of the five MHC DQB alleles were present in both regions, albeit in differing frequencies across their range. The only exception is one unique allele observed in a single Pacific walrus individual. Thus, it does not appear that varying environments experienced by Pacific and Atlantic walruses are influencing the strength of balancing selection exerted on DQB.
Implications for conservation
Future declines in summer sea ice are expected, perhaps reaching a state of a nearly ice-free summer before 2040 (Overland and Wang 2013). This reduction of sea ice will directly reduce the habitat available for ice-associated marine mammals, such as the walrus. In the North Pacific, summer sea ice is commonly utilized by female walruses and young calves, providing protection from terrestrial predators and a platform to rest near offshore foraging areas that are too far from land-based haul-out sites to be energetically feasible for feeding. With the increasing retreat of summer sea ice beyond the continental shelf, female walruses and their young have shifted their summer distribution and are utilizing coastal haul-out areas (Jay et al. 2012). These land-based haul-out areas can total over 100,000 individuals (Jay et al. 2011; MacCracken 2012), potentially increasing exposure of young walruses to pathogens associated with fecal matter and terrestrial systems along with other threats (e.g., predation, crushing, etc.; Laidre et al. 2008).
Arctic marine mammals may have a limited capacity for shifting their range northward. The continental shelf likely dictates the northernmost extent of the distribution for many species, such as the Pacific walrus, as the deep waters located past the shelf may not be optimal for foraging (Jay et al. 2012). Atlantic walruses may be more resilient to changes in sea ice conditions as they utilize more near-shore habitat for foraging (Born 2005). Regardless, expansion of temperate marine mammals into the Arctic, and changes in pelagic and benthic invertebrate faunas (prey) and environmental conditions may result in the transfer of pathogens to new areas and potentially immunologically naïve wildlife (Rausch et al. 2007; Kovacs et al. 2011). Predicted changes in Arctic ecosystems include the expansion of arthropod vectors into northern habitats and increased maritime traffic, likely resulting in increased pathogen transmission through direct and indirect contact with humans and their domesticated animals (Hueffer et al. 2011). Reductions in summer sea ice are already sufficient to enable the exchange of marine mammals through the Northwest Passage between the Atlantic and Pacific Oceans (i.e., bowhead whale, Balaena mysticetus; Heide-Jørgensen et al. 2012), facilitating the transfer of pathogens associated with conspecifics through secondary contact. Until sufficient gene flow homogenizes genetic diversity (and frequencies) at genes associated with immune response, populations of marine species coming into contact after long periods of isolation may be at increased risk to the transfer of novel pathogens and subsequent immune challenges.
Shifts in ecological communities coupled with environmental stressors resulting from climate changes may challenge the immunological response of Arctic populations, especially species that may have reduced capacity to respond to immunological challenges given limited genetic diversity at genes associated with immune response. Walruses residing in the Pacific Ocean and in waters off of eastern Greenland may be more susceptible to novel pathogens relative to their western and northwestern Greenland counterparts, because they are predominately represented by ≤ 2 alleles, and the two MHC alleles are identical at ABS. However, additional samples from the Atlantic populations are needed to further assess the levels of genetic diversity at genes associated with immune response. Study of additional genes associated with both adaptive (e.g., other genes within MHC) and innate (e.g., Toll-like receptors) immune response are needed to fully understand the ability of Arctic vertebrate populations, including walruses, to respond to increased exposure to novel pathogens.
This research was funded by the US Geological Survey, Ecosystems Mission Area, Wildlife Program. Research was conducted under institutional animal care and use committee approval (approval number 06SOP06). We thank the scientific staff and crew for sample collections during cruises aboard the USCG Healy, R/V Magadan, and R/V Stimson; E.W. Born for providing samples from Atlantic walruses; A. Fischbach for assisting with Pacific walrus sample collection; and R. Dial, Alaska Pacific University, for undergraduate mentorship of K.F. The manuscript was improved by comments from C. Lindqvist, University at Buffalo, and R. Wilson, University of Alaska Fairbanks. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement of the US Government.