, Volume 140, Issue 7, pp 349–364

MHC class I variation in a natural blue tit population (Cyanistes caeruleus)


    • Department of Behavioural Ecology and Evolutionary GeneticsMax Planck Institute for Ornithology
  • K. Foerster
    • Comparative ZoologyInstitute of Evolution and Ecology, University of Tübingen
  • B. Kempenaers
    • Department of Behavioural Ecology and Evolutionary GeneticsMax Planck Institute for Ornithology

DOI: 10.1007/s10709-012-9679-0

Cite this article as:
Wutzler, R., Foerster, K. & Kempenaers, B. Genetica (2012) 140: 349. doi:10.1007/s10709-012-9679-0


The major histocompatibility complex (MHC) is central to the vertebrate immune system and its highly polymorphic genes are considered to influence several life-history traits of individuals. To characterize the MHC in a natural population of blue tits (Cyanistes caeruleus) we investigated the class I exon 3 diversity of more than 900 individuals. We designed two pairs of motif-specific primers that reliably amplify independent subsets of MHC alleles. Applying denaturing gradient gel electrophoresis (DGGE) we obtained 48 independently inherited units of unique band patterns (DGGE-haplogroups), which were validated in a segregation analysis within 105 families. In a second approach, we extensively sequenced 6 unrelated individuals to confirm that DGGE-haplogroup composition reflects individual allelic variation. The highest number of different DGGE-haplogroups in a single individual corresponded in 19 MHC exon 3 sequences, suggesting a minimum of 10 amplified MHC class I loci in the blue tit. In total, we identified 50 unique functional and 3 non-functional sequences. Functional sequences showed high levels of recombination and strong positive selection in the antigen binding region, whereas nucleotide diversity was comparatively low in the range of all passerine species. Finally, in a phylogenetic comparison of passerine MHC class I exon 3 sequences we discuss conflicting evolutionary signals possibly due to recent gene duplication, recombination events and concerted evolution. Our results indicate that the described method is suitable to effectively explore the MHC diversity and its ecological impacts in blue tits in future studies.


MHC class IBlue titCyanistes caeruleusDGGEPopulation genetics


In the last decades, the major histocompatibility complex (MHC) has been the focus of many studies in molecular evolution and behavioural ecology (reviewed in Piertney and Oliver 2006). The MHC is of particular interest, because it determines an individual’s immunological capacity and is therefore linked to important aspects of life-history (e.g. survival and reproductive success; reviewed in Milinksi 2006). This makes the MHC genes an ideal candidate for studying selection operating in wild populations.

The essential function of the MHC is the initiation of the adaptive immune response via uptake of pathogen derived peptides and their presentation to T-cells (Klein 1986). There are two major types of MHC receptors (Murphy et al. 2008): Class I molecules are located on the surface of all nucleated somatic cells responsible for the display of intracellular antigens. Class II molecules are expressed only on antigen-presenting cells, which phagocytise extracellular pathogens and present the processed antigens to lymphocytes. The extraordinarily high level of polymorphism observed in MHC genes across a wide taxonomic range of vertebrates is supposed to be maintained by balancing selection (Potts and Wakeland 1993). This dynamic process is mainly driven by an evolutionary arms race between pathogens and their hosts, where pathogens escape the recognition ability of their hosts MHC genes and the hosts in turn evolve new MHC alleles to remain resistant (Apanius et al. 1997; Wegner et al. 2004). Alternatively, or complementarily to this disease-based force, MHC diversity is thought to be also maintained by mate choice (Milinski 2003; Ziegler et al. 2005). As each MHC molecule can bind to a specific spectrum of pathogenic antigens, individuals carrying rare alleles (“negative-frequency-dependent selection hypothesis”, Bodmer 1972) or exhibiting different alleles at certain MHC loci (“overdominance hypothesis”, Doherty and Zinkernagel 1975) benefit from enhanced disease resistance and thus enhanced survival. Individuals with rare or advantageous alleles at MHC loci may be additionally favoured by directional mating preferences and thus acquire increased reproductive success (“good genes hypothesis”, Møller 1999). Furthermore, individual MHC-based disassortative mating preferences result in maximised fitness by optimising the offspring’s MHC constitution (“compatible genes hypothesis”, Neff and Pitcher 2005).

The rapid evolution of MHC genes is proposed to be generated by various molecular mechanisms (Hughes et al. 2008). As shown by several studies, positive selection clearly contributes to MHC polymorphism by increasing the rate of non-synonymous nucleotide substitutions, mainly in sites encoding the antigen binding region (ABR). On the other hand, recombination and gene conversion events create new alleles, and hence are considered as another process how MHC diversity is increased. Gene duplication events further increase the number of genes, while some may get deleted and others retain in function or get inactivated and thus maintain in the population.

Information on MHC gene organization and empirical support for selection shaping MHC polymorphisms in wild populations are found extensively in studies of mammals (reviewed in Sommer 2005) and fish (Milinski 2003; Reusch et al. 2001). In birds, however, there is a comparable small number of studies on MHC diversity, which are mostly based on galliform species and a few passerines (e.g. Gillingham et al. 2009; Eimes et al. 2010; Freeman-Gallant et al. 2002). After the complete sequencing of the MHC class I and II genes (Kaufman et al. 1999), the domestic chicken became the model species for avian MHC. The so-called “minimal essential MHC” of the chicken is a small, densely packed gene region of exclusively functional genes located on chromosome 16 (Miller et al. 2004). This ‘simple’ structure was also found in other galliforms such as the ring-necked pheasant (Phasianus colchicus, Wittzell et al. 1999), black grouse (Tetrao tetrix, Strand et al. 2007), and in other avian groups like penguins (Sphenisciformes, Bollmer et al. 2007), owls (Strigiformes, Burri et al. 2008a, b), parrots (Psittaciformes, Hughes et al. 2008), subantarctic seabirds (Procellariiformes, Strandh et al. 2011) and ratites (Struthioniformes, Miller et al. 2011). Investigations on other galliform species such as the Japanese quail (Coturnix japonica, Shiina et al. 2004), the domestic turkey (Meleagris gallopavo, Chaves et al. 2009) and the greater prairie-chicken (Tympanuchus cupido, Eimes et al. 2010) revealed a more complex overall structure. Recent studies on birds of prey revealed that a different ‘simple’ MHC structure, which shows no pseudogenes and has a small number of loci, undergoes strong positive selection (e.g. Alcaide et al. 2009).

To date, the most complex avian MHC structure is found in passerines (Westerdahl 2007). Although there is a large variation in the MHC organization between the characterized species, most have a large number of genes, many gene copies, pseudogenes and long introns (e.g. Westerdahl et al. 2004; Anmarkrud et al. 2010). Furthermore, gene conversion events involving long fragments lead to homogenization of alleles from different loci, erasing locus-specific patterns (Hess and Edwards 2002; Ohta 1999).

Although technically challenging, there is an increasing number of studies describing the MHC class II organization in passerines (e.g. red-winged blackbird (Agelaius phoeniceus), Gasper et al. 2001; Hawaiian honeycreepers (Drepanidinae), Jarvi et al. 2004; New Zealand robin (Petroica australis), Miller and Lambert 2004; tanagers (Thraupidae), Sato et al. 2001; Darwin’s Finches (Geospizinae), Sato et al. 2011; starlings (Sturnidae), Wittzell et al. 1998). In contrast, MHC class I genes have been reported only for seven passerine species and were described in more detail in just three (house sparrow, Passer domesticus, Bonneaud et al. 2004b; scarlet rosefinch, Carpodacus erythrinus, Promerová et al. 2009; great reed warbler, Acrocephalus arundinaceus, Westerdahl et al. 1999). Recent studies indicate that MHC class I genes may be powerful immunogenetic markers for wild passerine populations (Bonneaud et al. 2006b; Loiseau et al. 2008; Westerdahl et al. 2005). For example, these studies found an association between specific MHC alleles and resistance to certain malaria parasites. Furthermore, only a few studies have investigated the role of passerine MHC class I genes in mate choice and sexual selection (Bonneaud et al. 2006a; Richardson et al. 2005; Westerdahl 2004).

The aim of this study was to develop and test a fast screening method to reliably quantify individual MHC class I diversity and assess variation within a natural population of the blue tit (Cyanistes caeruleus). Characterizing MHC genes in blue tits is particularly worthwhile. The blue tit is currently the subject of a number of long-term field projects in Europe and much is known of its ecology and behaviour (e.g., Doutrelant et al. 2000; Foerster et al. 2003; Kempenaers et al. 1992; Parker et al. 2006). Furthermore, blue tits are widely used as a model system for studies on sexual selection and mate choice (e.g. Andersson et al. 1998; Johnsen et al. 2003; Kempenaers et al. 1997), and several studies have highlighted the consequences of blood- and ecto-parasite burdens in blue tits (Arriero 2009; Bensch et al. 2000; Merino et al. 2000; Stjernman et al. 2004; Tomás et al. 2007; Tomás et al. 2006). In a first step towards the exploration of the blue tit MHC, Schut et al. (2011) recently detected 17 MHC class I alleles in 20 individuals originating from different European locations. By applying cloning-sequencing, their results indicate low levels of MHC class I variation in comparison to other passerine species.

In contrast, in this study, we used a multilocus motif-specific amplification in combination with denaturing gradient gel electrophoresis (DGGE) and cloning to characterize the MHC class I region of the blue tit. To effectively investigate the genetic variation within a single population, we screened more than 900 individuals via DGGE and verified MHC genotypes through segregation analysis. Additionally, we sequenced a subset of individuals representing the spectrum of DGGE genotypes of the population, to evaluate the performance of our method. We examined the effects of selection and recombination in maintaining MHC class I diversity within the obtained sequences. Finally, we conducted a phylogenetic comparison of the blue tit with available MHC class I sequences of other passerine species to gain further information on the evolution of passerine MHC class I genes.

Materials and methods

Study population and sampling

We collected samples and life-history data from a blue tit population located at Kolbeterberg in Vienna, Austria (48°13′N, 16°20′E). This population has been intensively studied since 1998 (e.g. Foerster et al. 2003, 2006; Poesel et al. 2001) and the study area encompasses 35 ha equipped with 250 nest boxes. Between 1998 and 2006, breeding birds and their offspring were captured and individually banded. Blood samples (10–100 μl) were obtained by brachial venipuncture, diluted in 800 μl of Queen’s lysis buffer (Seutin et al. 1991) or 100 % ethanol and stored at 4 °C for further analyses. Genomic DNA extractions were performed for all samples using the GFX Genomic Blood DNA Purification Kit (Amersham Biosciences, Freiburg, Germany).

Primer design

At first, we used genomic DNA (gDNA) of two adult blue tits (BT1 and BT2) caught in Vienna. To examine transcribed MHC sequences, RNA from the liver of a blue tit chick collected under license in Seewiesen, Germany (BT3) was extracted. mRNA was isolated using the Oligotex Direct mRNA Micro Kit (Qiagen, Hilden, Germany) and reverse transcription to coding DNA (cDNA) was performed with the 1st Strand cDNA Synthesis Kit (Roche Applied Science, Mannheim, Germany). We amplified gDNA and cDNA in 10 μl PCR reactions containing 5 ng DNA, 0.1 mM each dNTP, 2.5 mM MgCl2, 0.5 μM each primer, and 0.25 U Taq DNA polymerase in 1xPCR buffer (GeneAmp PCR Kit, Perkin Elmer, Rodgau, Germany). We ligated the resulting fragments into TOPO TA (Invitrogen, Darmstadt, Germany) and pGEM-T Easy (Promega, Mannheim, Germany) plasmid vectors, which were used to transform competent E. coli. We sequenced inserts from positive clones on an ABI PRISM 310 Genetic Analyser (Perkin Elmer, Rodgau, Germany) using the BigDye terminator sequencing kit (Perkin Elmer, Rodgau, Germany).

To design appropriate motif-specific primers, we first amplified a large part of exon 3 from gDNA, using the degenerate primers A21B and A23B (Bonneaud et al. 2004a, b, see Table 1). To sequence the entire exon 3, primer sites in exon 2 and exon 4 were targeted. We tried several known avian primers for these regions (Westerdahl et al. 1999, 2004), but only the combination of A21B and HN22 (Westerdahl et al. 1999, see Table 1) specifically amplified MHC fragments. The resulting sequences allowed us to design the primer PCA20 in exon 4. For a primer site in exon 2, we aligned published avian MHC sequences, and designed PCA10 in a conserved region near the 3′ end of this exon (Fig. 1). Then, we amplified, cloned and sequenced intron 2, exon 3, and intron 3 repeatedly from gDNA using the four primer combinations PCA10-A23B, PCA10-PCA20, A21B-A23B and A21B-PCA20. The reported gDNA sequences of exon 3 stem from clones of 11 ligated PCR products of BT1, and 11 ligated PCR products of BT2. The reported cDNA sequences stem from 10 ligated PCR products from BT3. These are verified sequences, i.e. sequences that were found at least in clones from two independently amplified and ligated PCR products.
Table 1

Degenerate primers used for amplification of MHC class I sequences in the blue tit

Primer name

Primer sequence in 5′→3′ direction

A21B (Bonneaud et al. 2004a, b)


A23B (Bonneaud et al. 2004a, b)


HN10 (Westerdahl et al. 1999)


HN22 (Westerdahl et al. 1999)













Fig. 1

Positions of primers used for the amplification of the blue tit MHC class I exon 3

In a second step, based on 11 known exon 3 sequences, we designed new primers near the ends of exon 3: PCA21 at the 3′ end and PCA11, PCA12, or PCA13 at the 5′ end, such that all known transcribed sequences could be amplified by one of these primer combinations (Fig. 1).

Sequence-specific amplification of MHC class I exon 3 and DGGE performance

The sequence-specific amplification with any one primer combination (PCA21 with PCA11, PCA12, or PCA13) ensured an amplification of a specific group of exon 3 sequences, which should optimally result in a suitable number of sequences to be separated using the DGGE method (Myers et al. 1987). Because melting temperature of the sequences was higher at the 5′ end then at the 3′ end of the fragment, we attached a GC-clamp to PCA11-13 as required for the DGGE.

When we optimized PCR and DGGE conditions, we were not successful in obtaining readable and repeatable band patterns for PCA12-PCA21; therefore, we used the two primer sets PCA11-PCA21 and PCA13-PCA21 and conducted 10 μl PCR reactions as described before. For DGGE, we produced gels containing 7 % 19:1 acrylamide/bisacrylamide, 1xTAE, formamide, and a 30–65 % denaturing gradient of urea for PCA11-PCA21, and 40–70 % urea for PCA13-PCA21. Gels were run at 60 °C in 1xTAE for 16.5 h at 180 V and always included two standard lanes with a mixture of amplified fragments from previously typed individuals. Gels were stained with SYBRgold (Molecular Probes, Darmstadt, Germany) to visualize the fragments under UV light. We followed this protocol for BT1-3 and a subset of 50 other blue tits, where we excised a random sample of 60 DGGE-bands. We re-amplified and sequenced the extracted DNA to verify that it only contained exon 3 sequences.

Assessing variation in the population

We used the DGGE method to separate MHC class I exon 3 alleles and visualize individual’s MHC diversity. We screened our study population according to unique and highly repeatable DGGE-band patterns obtained with both primer combinations (PCA11-PCA21 and PCA13-PCA21). For MHC genotyping we used DNA samples from 499 adult blue tits and 440 offspring. In total, we performed PCRs and DGGE gels from 918 individuals. Each amplified allele optimally results in a single DGGE-band, as its nucleotide composition determines migration distance. However, misinterpretation of bands on DGGE gels can lead to over- or underestimation of the individual’s MHC diversity. Similar sequences may migrate similar or equal distances on gels and may therefore reduce the number of bands, whereas heteroduplex formation results in an enhanced number of DGGE-bands (Westerdahl et al. 2004). From sequencing various excised DGGE-bands, we detected sequences which appeared as distinct DGGE-bands within one individual (Tamura et al. 2007). A segregation analysis of DGGE-bands allowed us to confirm their Mendelian segregation. Within our population, 105 families were genotyped for both primer combinations. Molecular analysis using microsatellites determined paternity (Foerster et al. 2003). Our sample included 77 males, 80 females (some breeding over several years) and 440 chicks; the number of offspring in each clutch varied between one and 13. In 29 cases, we could follow the segregation of DGGE-bands over three generations. DGGE-bands were transmitted as units (defined as DGGE-haplogroups), which were unique and consistent between individuals.

We excluded the possibility of cross-amplification of identical exon 3 alleles with both primer sets by performing a Fisher’s Exact Test with Bonferroni correction including all DGGE-haplogroups. By comparing the frequency of all possible DGGE-haplogroup combinations that could potentially occur if linked, we could exclude any over-representation shown within the population (p > 0.05).

Investigation of DGGE-haplogroup composition

The information on exon 3 sequences from BT1, BT2, and BT3, as well as the DGGE genotypes provided only a rough estimate of the individual allelic diversity at the MHC class I loci. To obtain a more accurate count of amplified MHC alleles for single individuals, and to further investigate whether DGGE-haplogroups actually represent an individual’s diversity, we cloned and sequenced (as described in the ‘Primer Design’ section) the PCR products of three unrelated individuals with low (BT4, N = 3), intermediate (BT5, N = 6) and high (BT6, N = 12) numbers of DGGE-haplogroups. For each individual, we conducted several PCRs (BT4: 11 and 9, BT5: 7 and 7, BT6: 7 and 5; for PCA11-PCA21 and PCA13-PCA21, respectively). We randomly selected 1–21 positive clones from each PCR for sequencing (MWG Biotech AG, Ebersberg, Germany), and sequenced a total of 100 positive clones per individual (50 for each primer combination). We report only verified sequences, i.e. sequences that were found in at least two clones from two independently amplified and ligated PCR products. To further prove the uniqueness of our highly similar MHC I sequences, we performed a screening using the program Mallard (Ashelford et al. 2006). This tool is based on the Pintail algorithm and detects possible anomalous sequences and chimeras within a multiple alignment (Ashelford et al. 2005). Using the default and most sensitive settings to decrease the number of false positives (cut-off line 99.9 %) not one single outlier was identified (N = 50, DEdiff = 0). We therefore infer that our perceived sequence diversity is not biased. These newly obtained MHC I exon 3 sequences were added to the NCBI Genbank with accession numbers HQ393911-HQ393951. To better distinguish them from previously published sequences (paca-UA*1-13, Foerster et al. unpublished; and paca-UA*101-117, Schut et al. 2011), we refer to these sequences using the abbreviation cyca-UA*14-53 (referenced to Cyanistes caeruleus).

Sequence analysis

We performed all alignments and translations into amino acid sequences in Bioedit 5.0.9 (Hall 1999). As no chimeras or other anomalous sequences were detected within our set of MHC I sequences, we assumed that the high rates of recombination lead to highly similar sequences. We used the programs DnaSP 5.10.01 (Rozas 1999) and Geneconv 1.81 (Sawyer 1989) to identify possible recombination sites and estimate the minimum number of inferred recombination events (RM, Hudson and Kaplan 1985). The nucleotide diversities (π) within all functional sequences and between BT4, BT5 and BT6 were calculated using DnaSP 5.10.01.

Evidence for selection was assessed through various approaches. To detect whether positive selection (ω(dN/dS) > 1) or purifying selection (ω(dN/dS) < 1) is acting on specific sites within the MHC class I exon 3 sequences we used Datamonkey (Pond and Frost 2005a). The models implemented in the Hyphy package (Pond et al. 2005) tested a variety of molecular evolutionary hypotheses based on maximum likelihood analyses. Firstly, selection was estimated using the Random Effects Likelihood (REL) model, which performs best for small or low divergence alignments (Pond and Frost 2005b). This model uses the empirical Bayes approach at each codon site, which allows for synonymous rate variation according to a predefined distribution. Secondly, we applied the Fixed Effects Likelihood (FEL) model, which directly identifies rates of non-synonymous and synonymous substitutions at each codon. Additionally, we used the random-site models of CodeML implemented in the package PAML 4 (Yang 1997, 2007). For each substitution rate category, the dN/dS ratio (ω) and the proportion of codon sites that fall into that category are determined from the data. Positively selected codons were identified using the Bayes empirical Bayes (BEB) approach, implemented in the site models M2a and M8 (Yang et al. 2005). M2a and M8 both allow for positive selection, while M2a allows values of ω > 1 and M8 adds an extra class of sites with 0 < ω < 1, ω > 1 with ω to vary among sites according to a beta distribution. The detected exon 3 sites were compared to those found in other passerines and chicken. In addition to these site-by-site selection detection approaches, we calculated overall values of dN/dS, both within ABR and non-ABR, using the Nei-Gojobori method (Mega 4.0, Tamura et al. 2007) and Tajima’s D as a supporting indicator for selection (DnaSP). We defined the ABR and non-ABR in accordance to great reed warbler MHC class I sequences (Westerdahl et al. 1999).


A phylogeny of the obtained sequences from individuals BT4-6 and previously published blue tit sequences was inferred using PhyML 3.0 (Guindon and Gascuel 2008). To achieve the tree topology, we performed best of NNIs (Nearest Neighbor Interchange) and SPR (Subtree Pruning and Regrafting) as type of tree improvement. We selected the GTR substitution model after justification by Akaike information criterion in JModeltest (Posada 2009) and calculated bootstrap support for nodes using 1000 replications.

Moreover, we performed a phylogenetic analyses on the evolutionary relationships of the MHC class I exon 3 alleles among passerine species. As conflicting signals such as recombination or duplication events might be expected to occur in MHC genes, we built a phylogenetic network rather than rely on a single traditional phylogenetic tree. A distance matrix built with the JC model was used in a Neighbor-Net analysis (Bryant and Moulton 2004) as implemented in Splitstree4 (Huson and Bryant 2006). The network shows all available Paridae MHC class I sequences (house sparrow, Bonneaud et al. 2004b; Loiseau et al. 2008; blue tit, Foerster et al. unpublished and Schut et al. 2011; great tit, Binz et al. unpublished and green-backed tit, Lo et al. unpublished). However, to simplify the network and focus on the Paridae cluster, we only added one randomly chosen representative sequence from each of the other passerine species (scarlet rosefinch, Promerová et al. 2009; seychelles warbler, Richardson and Westerdahl 2003; great reed warbler, Westerdahl et al. 2004). Analyses including all available sequences provided similar results.


Characterization of blue tit MHC class I genes

In an initial screening based on clones from three individuals (BT1-3), 9 functional and 2 non-functional exon 3 sequences (paca-UA*11, paca-UA*12) were detected. Moreover, BT1-3 sequences from excised DGGE bands matched those from cloned cDNA and gDNA fragments, and we found all but 3 of the previously cloned sequences on the DGGE gel. In addition we obtained 55 readable MHC class I exon 3 sequences from 60 randomly excised DGGE bands of a subset of 50 unrelated birds. In summary, using the primer combinations PCA11-PCA21 and PCA13-PCA21 (Fig. 1) the DGGE analysis confirmed eight sequences and detected one additional sequence (paca-UA*10) amongst BT1-3 and 50 unrelated individuals. Earlier these MHC I exon 3 sequences were published in NCBI Genbank with accession numbers AM232705-AM232717.

Clones from three additional birds (BT4-6) resulted in 41 additional sequences of which only one was non-functional (cyca-UA*13). We therefore identified 53 sequences in total (Online Resource 1), which varied in length between 207 and 276 bp, covering the major part of the MHC class I exon 3 (depending on the primer combination, Online Resource 2). Three of these verified sequences were apparent pseudogenes, since they contained one or more stop codons (Online Resource 1).

Translated amino acid sequences were aligned to other MHC class I exon 3 sequences (chicken, Guillemot et al. 1988; great reed warbler, Westerdahl et al. 1999, Fig. 2). Overall, the 50 functional blue tit MHC class I exon 3 sequences showed 33 polymorphic sites on a common sequence length of 165 bp, and had a nucleotide diversity of π = 0.072 (Table 2). We found no indication of balancing selection across all sites (dN/dS = 1.267, Z = 0.602, p = 0.522), but a high, significant value of Tajima’s D (D = 2.069*), which suggests that alleles must have been maintained for longer than expected under neutrality. In contrast to the non-ABR region, where moderate negative selection was detected (dN/dS = 0.770, p = 0.677, Z = −0.417), the ABR appears to be under strong balancing selection (dN/dS = 3.821, p = 0.006, Z = 2.528*) and sites seem to be fixed (D = 2.528*).
Fig. 2

Amino acid sequences of all 50 functional blue tit MHC class I exon 3 sequences compared to other blue tit (paca-UA*101-117, Schut et al. 2011), great reed warbler (A. arunicainsCAA06560, Westerdahl et al. 1999), and chicken (CAA31272, Guillemot et al. 1998). Position 1 of the alignment corresponds to the first position of the α2-domain. Identity with the sequence paca-UA*2 is indicated by dots (·), length differences without information (as a result from the use of different primers) by a squiggle (~), and inserted gaps to obtain the best alignment by dashes (–). Shaded columns represent codons for putative ABRs and highly conserved sites are in rectangles as suggested from human and chicken sequences (Koch et al. 2007; Wallny et al. 2006). Positions under positive selection (by FEL, REL and CodeML) in the common sequence length of 55 codons are indicated by a plus sign (+). Sites under negative selection in the blue tit are depicted as hash (#). Positively selected sites (as identified in Promerová et al. 2009) in the great reed warbler sequence (Westerdahl et al. 1999) are printed on black circles

Table 2

Selection indices and overview about ω(dN/dS) ratio within all blue tit MHC I sequences (paca-UA*1-10 and cyca-UA*13-53), group PCA11-21, group PCA13-21 with ABR and non-ABR


All sequences

Group PCA11-PCA21

Group PCA13-PCA21

All sites







Tajima’s D








n ± SD

0.072 ± 0.00

0.184 ± 0.01

0.050 ± 0.03

0.044 ± 0.05

0.032 ± 0.02

0.162 ± 0.09

0.041 ± 0.02

































dN ± SE

0.076 ± 0.02

0.214 ± 0.07

0.047 ± 0.01

0.048 ± 0.04

0.023 ± 0.01

0.185 ± 0.05

0.042 ± 0.01

dS ± SE

0.060 ± 0.02

0.056 ± 0.05

0.061 ± 0.02

0.039 ± 0.03

0.061 ± 0.03

0.073 ± 0.06

0.036 ± 0.02

























Tajima’s D values and the significance to the assumption under neutrality (p < 0.05 indicated with *), π (nucleotide diversity), m (number of sequences), S (number of segregating sites), N (number of sites of the analyzed region), ps (the number of segregating sites divided by the total number of sites) and Z (codon based Z-test, Nei-Gogobori, Jukes Cantor)

Before we investigated whether selection ω(dN/dS) was acting on specific codons, we tested for recombination within the alignment of 50 functional MHC I blue tit sequences. The program DnaSP identified a minimum number of recombination events (RM) of 7 (8/52, 56/60, 68/114, 114/119, 119/127, 127/140, 143/156), whereas GeneConv found evidence for putative recombination events at 2 sites (8/121, 9/97).

Among the alignment of 55 codons, the REL model identified 11 sites with evidence for positive selection (dN > dS, positions 23, 31, 38, 48, 49, 57, 58, 63, 66, 71, 76), whereas the FEL model detected only 2 (dN > dS, positions 23, 63). Both models reported the same 8 codons under negative selection (dN < dS, positions 27, 28, 33, 37, 43, 61, 72, 75). Models M2a and M8 implemented in CodeML identified the same 5 positively selected sites with posterior probabilities greater than 95 % in Bayes empirical analysis (23, 25, 63, 66, 71). When considering only codons assigned by at least two analyses, we found 4 positively (positions 23, 63, and 66 were detected three- and 71 two times) and eight negatively selected sites (only REL and FEL; Fig. 2).

A phylogenetic tree including previously published sequences by Schut et al. (2011) revealed two major clusters of functional sequences supported by a bootstrap value of 730. This can be associated with the sequences amplified by the two distinct primer sets used (Fig. 3). This suggests that the primer combinations PCA11-PCA21 and PCA13-PCA21 may amplify two distinct blue tit sequence groups (Table 2) from multiple MHC class I loci. The three pseudogenes differed clearly (probability value of 100) from all other functional sequences. Sequences, which were described earlier (Schut et al. 2011) were spread amongst our sequences and were present in both groups. However, these sequences did not cluster in the way it was stated by Schut et al. (2011). Yet, similar to Schut et al. (2011), these clusters were significantly different in their nucleotide diversity (group PCA11-PCA21: π = 0.034 ± SD 0.006, S = 20; group PCA13-PCA21: π = 0.061 ± SD 0.004, S = 26; p < 0.01, t = 20.11) and selection in ABR and non-ABR regions (Table 2). The group PCA11-PCA21 showed evidence for purifying selection in the non-ABR region (dN/dS = 0.377, Z = −1.523, p = 0.130) and neutral variation in the ABR (dN/dS = 1.231, Z = 0.660, p = 0.510). In contrast, group PCA13-PCA21 showed evidence for positive selection (dN/dS = 2.534, Z = 2.067, p = 0.04) and a significant value of Tajima’s D = 2.409* in the ABR region. The non-ABR of the PCA13-PCA21 sequences did not show evidence for selection (dN/dS = 1.167, Z = 0.215, p = 0.82).
Fig. 3

Phylogenetic tree for all available blue tit MHC class I exon 3 sequences (paca-UA*1-11; paca-UA*101-117 and cyca-UA*13-53), with bootstrap support for nodes (1000 replications). Topology presented is the best of best of NNIs and SPR under the GTR model. Solid dots represent sequences amplified with primer combination PCA 13-21, while open dots indicate sequences amplified with PCA 11–21. Numbers in circles correspond to sequences obtained from individuals BT4-6. The scale bar indicates distance in units of nucleotide substitutions per site

MHC class I variation in the population

We screened 918 individuals (breeding adults and chicks) for both primer sets. DGGE-bands were in number and composition repeatable within individuals and unique between individuals. Several DGGE-bands (1 to 7) were linked as units and to confirm these units (DGGE-haplogroups), we followed their transmission from parents to chicks within 105 families. Bands that were not verified in DGGE-haplogroups in the segregation analysis were excluded from further analyses (e.g., bands that appeared in chicks but not in parents, or vice versa). We found such bands in 6 and 8 % of the typed families for the primer combination PCA11-PCA21 and PCA13-PCA21, respectively. We identified 30 separately inherited unique DGGE-haplogroups with the primer combination PCA11-PCA21 and 18 with PCA13-PCA21. Note that DGGE-haplogroups are not true haplotypes, since more than two haplogroups were present in one individual. Due to the overlapping primer sets, we tested for linkage between the 48 DGGE-haplogroups to exclude possible cross-amplification. None of the DGGE-haplogroups were linked (Fisher’s Exact Test with Bonferroni correction, with 540 possible combinations within the population N = 918, p > 0.05). This suggests that our two primer sets amplify independent, unique DGGE-haplogroups. Using this simplified mapping of DGGE-haplogroups we screened 499 breeding adults for MHC variation. The average number of DGGE-haplogroups was four and individual diversity ranged between two and 12 DGGE-haplogroups (Online Resource 3).

Evaluation of DGGE-haplogroups as a measure of sequence variation

To further evaluate DGGE-haplogroups, we compared the number of DGGE-haplogroups of three blue tits (BT4-6) with the number of distinct sequences obtained after cloning (see Table 3; Fig. 4). For BT4 and BT5, with 3 and 6 DGGE-haplogroups, respectively, we identified 12 sequences per individual. Nineteen sequences were found for the most diverse individual BT6, which showed 12 DGGE-haplogroups. BT6 shared one identical sequence with BT5 (cyca-UA*21) and one with BT4 (cyca-UA*28) (Table 3). Moreover, BT6 possessed a sequence (cyca-UA*41) identical to a sequence found in other individuals from different European populations (paca-UA*104). However, there was no shared sequence between BT1-3 and BT4-6, which is probably due to the less excessive sequencing of BT1-3 and the small number of individuals. When comparing nucleotide diversity of the three individuals, we found significant differences between BT4 and BT5 (p = 0.004, t = 3.04,) and between BT4 and BT6 (p = 0.006, t = 2.81), but not between BT5 and BT6 (p = 0.402, t = 0.84). The MHC alleles characterized through cloning/sequencing of BT4-6 individuals appears to detect higher number of MHC class I loci than the DGGE analysis. According to our sequence data, blue tits may have a minimum of 10 MHC class I loci.
Table 3

Number of DGGE bands, haplogroups and sequences of MHC class I exon 3 sequences obtained from three individual blue tits (BT4-6) using primer sets PCA11-PCA21 and PCA13-PCA21

The total number of DGGE bands, haplogroups and sequences detected in each individual and their corresponding nucleotide diversity (π ± SD) are shown. Asterisks refer to significance of T test between individuals (p < 0.05)
Fig. 4

MHC genotypes of three blue tits (BT4-6). DGGE-bands are aligned to marker lanes (M), where a mix of known genotypes was loaded. Black arrows indicate shared DGGE-haplogroups which potentially contain the sequences cyca-UA*28 and cyca-UA*21 in individuals BT4/BT6 and BT5/BT6, respectively. The non-functional sequence paca-UA*11 (ψ) was present on DGGE gels of all individuals. Stars (*) indicate double bands

MHC class I passerine phylogeny

A Neighbor-Net revealed structuring among passerine MHC class I exon 3 sequences (Fig. 5). All 53 verified blue tit MHC class I exon 3 sequences (including previously published blue tit sequences from Schut et al. 2011) were compared to those from other passerine species represented by one randomly chosen sequence (see Methods) and Paridae sequences from the great tit (Binz et al. unpublished) and the green-backed tit (Lo et al. unpublished). The topology of the Neighbor-Net showed numerous conflicting signals, suggesting extensive recombination events within the blue tit, but also between species. The Paridae form a divergent cluster from the other three passerine sequences. Within the Paridae, the blue tit sequences formed a clade with many subsets, which may correspond to different loci. One great tit and two green-backed tit sequences (pama*7, pamo*7 and pamo*11) intermingled in one large blue tit subset amplified by primer combination PCA13-PCA21, which may represent a paraphyletic group. All three blue tit pseudogenes clustered with great tit sequences pamo*1 and pamo*6, and separate from other functional sequences. These pseudogenes may have been maintained as trans-species polymorphism from a common ancestor of blue and great tits. The Neighbor-Net further revealed that newly obtained sequences and previously published sequences from other blue tit populations are intermingled, although they clustered in groups according to the primer set used. However, since more than just two major clusters become obvious (compared to Fig. 5), we cannot clearly say, if primer specific grouping is occurring and if only two MHC class I loci are amplified from MHC-B and MHC-Y, respectively.
Fig. 5

Neighbor-Net with Jukes-Cantor model as implemented in Splitstree4 was used to generate the distance matrix with blue tit MHC class I exon 3 sequences in relationship to all available Paridae exon 3 sequences and one randomly chosen sequence of passerine species (house sparrow, P. domesticus, pado EF429132, Bonneaud et al. 2004a, b; Loiseau et al. 2008; scarlet rose finch, C. Erythrinus, caer FJ392762, Promerová et al. 2009; seychelles warbler, A. Sechellensis, acse AJ557874, Richardson and Westerdahl 2003; great reed warbler, A. arundinaceus, acar AJ005503, Westerdahl et al. 2004). Colorful labels depict Paridae sequence origin (blue tit, C. caeruleus, paca-UA*1-12 in light blue, cyca-UA*13-53 in dark blue, paca-UA*101-117 in lilac; great tit, P. major, pama in orange, Binz et al. unpublished; green-backed tit, P. monticolus, pamo in green, Lo et al. unpublished). The scale bar indicates distance in units of nucleotide substitutions per site. Parallel edges represent conflicting signal in the data set


Variation of blue tit MHC class I genes at the individual and population levels

We developed two pairs of motif-specific primers (PCA11-PCA21 and PCA13-PCA21) for the amplification of MHC class I exon 3 in the blue tit. The variation of individual MHC genotypes within the population was visualized using DGGE. DGGE is considered an extremely sensitive and rapid method for MHC typing, suitable for examining nucleotide sequence variation in large-scale population surveys (Miller et al. 1999; Langefors et al. 2000). In contrast to other studies in birds, we did not assume a priori that each amplified allele would result in one single DGGE-band (Bonneaud et al. 2004a, 2006b). We rather suspected that one to seven DGGE-bands detected consistently between individuals would be inherited as unique haplogroup units. Our segregation analysis within 105 families confirmed 48 such units, which were termed DGGE-haplogroups.

The two primer combinations used in this study shared one primer, but amplified sets of 30 and 18 DGGE-haplogroups, respectively. Individual MHC genotypes were composed of 2 to 12 DGGE-haplogroups, while the population average was 4. Apparently, the defined DGGE-haplogroups were unlinked; however, we cannot exclude the possibility that the limited power of the analysis prevented linkage detection. In a study of great reed warblers, tightly linked DGGE-bands were assigned to specific haplotypes, suggesting that MHC class I loci in passerines are located on a single chromosome (Westerdahl et al. 2004). This was also hypothesized by Balakrishnan et al. (2010), who suggested that zebra finch (Taeniopygia guttata) MHC class I loci may be located on chromosome 22. However, Ekblom et al. (2011) recently mapped zebra finch MHC class I genes to chromosome 16, as found in the chicken and other galliformes.

Another remarkable feature of the chicken MHC is that it consists of two independent clusters of classical MHC (MHC-B) and non-classical MHC (MHC-Y or Rfp-Y) genes. These linkage groups are known from many galliform species to be located on one chromosome, but are separated by a region of very high recombination. The MHC-Y genes have been described to be functional, but seem to be less polymorphic and might show lower expression than the MHC-B (e.g. Strand et al. 2007; Delany et al. 2009; Eimes et al. 2010). The results from this study suggest, that our primers might amplify both classical and non-classical MHC class I gene groups in the blue tit. However, it is unknown whether there is an MHC-Y locus in passerines, although clusters of MHC alleles with low variability have been previously reported (Jarvi et al. 2004; Bonneaud et al. 2004a, b).

The thorough cloning and sequencing of three individuals (BT4-6) differing in their DGGE-haplogroup number provided specific information about the composition of DGGE-haplogroups. Cloning and sequencing of the most diverse individual BT6, which possessed 12 DGGE-haplogroups, revealed 19 different sequences. For BT5 and BT4 (with 6 and 3 DGGE-haplogroups, respectively), we found 12 sequences each. The phylogeny of these sequences did not result in identification of specific loci; however, sequences from BT5 and BT6 grouped in a few small sub-clusters. In contrast, the 9 sequences of BT4 amplified by the primer combination PCA13-PCA21 clustered in two tight groups and formed two distinct DGGE-bands on gels.

As expected, BT4 had the lowest nucleotide diversity when compared to BT5 and BT6. This suggests that DGGE may underestimate diversity, since similar sequences may migrate similar distances on the gel. This illustrates the first potential conformation of a haplogroup, which may result from allele variants at the same locus or at closely linked loci. A second type of DGGE-haplogroups may not contain true alleles, possibly a result of heteroduplexes or chimeras, which can be formed through a specific combination of alleles, and which may appear as extra bands on the gel. It is likely that the DGGE-haplogroups reported in this study contained some heteroduplexes or chimeras formed by alleles from linked loci. By evaluating the number of haplogroups, rather than the number of bands, we avoided artificially increasing individual estimates of allelic diversity. If loci are not inherited together, their alleles could still form heteroduplexes or chimeras, but these would not appear as a heritable pattern on the DGGE gel. We detected some DGGE bands that showed non-Mendelian patterns in the pedigree analysis and we deleted them from the analysis. This has also been a common problem in other studies using motif-specific amplification to explore passerine MHC (Bonneaud et al. 2004b; Westerdahl et al. 2004).

The high similarity of alleles between loci is most likely the result of homogenisation through gene conversion. This process involves the transmission of long fragments from one gene to another, thus, erasing locus-specific patterns of alleles (Hess and Edwards 2002). As extensive gene duplication events are likely to have shaped the MHC in birds, this could contribute to a high similarity of alleles between loci. The number of DGGE-haplogroups reported is therefore a conservative lower estimate of the number of independently inherited MHC class I alleles of an individual blue tit. The varying number of DGGE-haplogroups between individuals probably reflects individual variation in heterozygosity/diversity, rather than variation in the number of MHC class I genes.

Our DGGE analysis revealed up to 12 haplogroups in one individual (BT6), suggesting at least the presence of 6 MHC class I loci in the blue tit. This is in agreement with the results of Schut et al. (2011), using restriction fragment length polymorphism (RFLP) analysis for visualization of alleles. However, 19 different MHC class I sequences were present in this individual (BT6), indicating at least 10 amplified MHC class I loci in the blue tit. Our sequence-based estimate derives from careful amplification and cloning, followed by conservative analyses to exclude false, chimeric or other anomalous alleles. The estimated number of blue tit MHC class I loci (with 6 to 10 loci) falls within the estimates for other passerine species. Scarlet rosefinches revealed a minimum of five MHC class I loci (Promerová et al. 2009), whereas great reed warbler and seychelles warbler exhibit at least 8 loci (Richardson and Westerdahl 2003; Westerdahl et al. 1999, 2004), and Bonneaud et al. (2004a, b) estimated between three and six MHC class I loci in the house sparrow.

Moreover, we were able to identify three MHC pseudogenes in the blue tit, one of them detected on DGGE gels in all individuals of the analyzed population. Generally indicated by stop codons or disrupted open reading frames, these non-functional genes are believed to be free of selective forces and thus can persist in the genome (Nei and Rooney 2005). Their phylogenetic position is therefore distinct to functional sequences. Pseudogenes have been commonly documented in passerines for MHC class II, but so far the occurrence of MHC classes I pseudogenes in passerines had only been reported in the great reed warbler and the zebra finch (Balakrishnan et al. 2010; Westerdahl 2007).

Here, we present a rapid MHC genotyping method that is sensitive enough to detect a representative minimum set of alleles to assess individual MHC class I variation in the blue tit. This method, although it cannot guarantee complete detection of MHC alleles as seen with total sequencing, still allows targeting variation in heritable MHC units defined as DGGE-haplogroups. As such, it may serve as a molecular tool to further address fundamental MHC-based questions within or between blue tit populations.

Selection and recombination in blue tit MHC class I genes

The main force maintaining MHC polymorphism seems to be balancing selection (Bernatchez and Landry 2003; Hedrick 1999). At sites under selection, an increased rate of non-synonymous substitutions leads to a permanent change in the amino acid sequence and, in the case of MHC genes, to a functional switch in the MHC protein. The intensity for such indications of adaptive processes due to parasite-driven selection should be highest at regions directly involved in binding foreign antigens (Potts and Wakeland 1990). Particular codons involved in building the ABR show signs of positive selection, whereas codons responsible for the preservation of the three-dimensional structure of the carrier protein are highly conserved and under purifying selection (Garrigan and Hedrick 2003). The analysed exon 3 of MHC class I sequences reported here contains the ABR in passerines (e.g. Westerdahl et al. 1999). Within our 50 functional exon 3 sequences, we found evidence for strong positive selection in the ABR (ω > 3), whereas the non-ABR seems to underlie moderate negative selection (ω < 1). A significantly positive value of Tajima’s D further confirmed their presence in the population for longer than would be expected under neutrality.

It has been suggested that recombination and balancing selection play an important role in avian MHC gene evolution (Edwards and Dillon 2004). Although recombination was not tested in the previous blue tit study of Schut et al. (2011), recombination was responsible for generating highly variable MHC class I exon 3 variants in the scarlet rosefinch (Promerová et al. 2009). Our study is consistent with the idea that recombination plays an important role in MHC diversity, as we were able to detect evidence for recombination events at several sites. Additionally, we found evidence for balancing selection at different codons. We suggest that these 4 sites are crucial for antigen recognition in the blue tit’s MHC class I, as 2 of them are matching sites defined in the human and chicken ABR regions (Koch et al. 2007; Wallny et al. 2006). We further detected 8 codons under purifying selection with FEL and REL (dN < dS), suggesting structural conservation at these residues (Kaufman et al. 1994; Mesa et al. 2004). There was only one site (position 32) that was found to be under opposite selection pressures. Taken together, the indices of positive selection, the nucleotide diversity and the number of segregating sites in our blue tit MHC I sequences were higher than those reported by Schut et al. (2011). Nevertheless, we confirm the variation of MHC class I genes in blue tits as low, compared to other passerine species (Schut et al. 2011). However, we found only 6 sequences being expressed from one individual. Therefore, further investigations are required to prove a correlation between signatures for balancing selection and expression rate.

Given the phylogenetic tree comparing all available MHC class I sequences, we can confirm two major MHC class I clusters in the blue tit. These two clusters, which differed significantly in nucleotide diversity, were detected by two different primer combinations. Group PCA11-PCA21 showed signs of purifying selection, whereas group PCA13-PCA21 showed strong positive selection acting on the ABR and non-ABR. The different types of selection mechanism operating in these groups of genes may imply different functional evolution. The high degree of sequence conservation and lack of polymorphism found in cluster PCA11-PCA21 are characteristic for the non-classical MHC-Y genes in galliformes (Eimes et al. 2010). We assume this cluster may eventually represent the counterpart to the MHC-Y genes in the chicken (Delany et al. 2009). These observations are consistent with findings from other passerines (Jarvi et al. 2004; Bonneaud et al. 2004a, b). However, more information on passerine MHC structure is required to confirm this.

Passerine phylogeny

The phylogeny of passerine MHC class I exon 3 genes in a Neighbor-Net suggests that the blue tit, green-backed tit and great tit form a clade separate from warblers, scarlet rosefinch and house sparrow sequences. This divergence reflects their taxonomic relationships as they do not belong to the family of Paridae (Johansson et al. 2008). As expected from the nature of the avian MHC, the Neighbor-Net showed conflicting evolutionary signals due to recent gene duplication, recombination events and/or concerted evolution within the Paridae clade (Hess and Edwards 2002; Westerdahl 2007; Burri et al. 2008a). Within the cluster of the Paridae, we found no clear species-specific clustering, although green-backed tit and great tit sequences were more similar to each other than to blue tit sequences. Based on analyses of several neutral markers, the great tit and the green-backed tit are closely related species of the genus Parus (Sheldon and Gill 1996; Slikas et al. 1996), whereas the blue tit belongs to the genus Cyanistes (Gill et al. 2005). Therefore, we did not find a strict monophyletic clustering by species, as would be expected under high rates of concerted evolution or recent gene duplications in highly divergent species (Edwards et al. 1995, Ekblom et al. 2003).

Some green-backed tit sequences clustered with great tit sequences; two of those grouped specifically with the three blue tit pseudogenes. We assume these pseudogenes originated prior to speciation events within the Paridae. Furthermore, the non-specific grouping of MHC sequences from the three species studied here also suggests the presence of ancestral polymorphisms that originated prior to the divergence of these species, and that were likely maintained through balancing selection (Klein 1987; Hedrick 1999; Bollmer et al. 2010).

The overall pattern of blue tit MHC sequences, composed of several sub-clusters, some of which can be grouped on the basis of primer specificity, suggests these sequences may represent allele variants from multiple MHC loci.

The molecular characterization of the passerine MHC class I gene family is in its infancy. Our study presents a new multilocus typing approach to assess MHC class I diversity in the blue tit. This approach can be useful for studies on population structure, host-pathogen evolution, and mate choice. Furthermore, the described technique can provide future insights into the structural organization of the passerine MHC.


We thank K. Persson, K. Teltscher, U. Holter and S. Kuhn for their excellent laboratory work and to all the field assistants who were involved in the Kolbeterberg project. We are grateful to H. Westerdahl for her help in designing the optimal primers and her support throughout the study. Furthermore we are indebted to C. Oppelt, for his inspiring comments on the manuscript. The present study was funded by the German Research foundation (DFG KE 867/1) and the Max Planck Society.

Conflict of interest

The authors declare that they have no conflict of interest.

Supplementary material

10709_2012_9679_MOESM1_ESM.doc (212 kb)
Supplementary material 1 (DOC 211 kb)

Copyright information

© Springer Science+Business Media B.V. 2012