Ethics statement

This study was approved by the University of Utah Institutional Animal Care and Use Committee, protocols 09-04015 and 10-05007. Private owners of birds signed a consent form allowing us to use blood and feathers in this study. Importation of feathers from outside the USA was approved under USDA APHIS permit 110106, issued to M.D. Shapiro. No material was collected from the wild for this study.

Discovery of variant alleles

We collected blood and feather samples and extracted DNA as previously described [27]. We selected a panel of 120 birds with diverse pigment phenotypes (recorded in photographs by us or as reported by breeders who donated feather samples via mail) from 68 breeds to characterize polymorphisms in the coding sequence of Mc1r. We focused on well-described phenotypes [23] that could be easily classified as blue (dark grey/black, the ancestral color morph), ash-red, brown, or modified versions of these colors such as their dilute forms (e.g., silver/dun, ash-yellow, and khaki, respectively). We also included recessive red and white birds.

Most of the Mc1r coding sequence was amplified using the avian degenerate primers MSHR72 (5’ ATGCCAGTGAGGGCAACCA-3’) and MSHR9 (5’-CTGGCT CCGGAAGGCATAGAT-3’) [6]; however, this sequence excluded the 5’ and 3’ ends of the coding sequence. We amplified the 5’ region of Mc1r containing the start codon using the primers mc1r_F2 (5’-CTTTAAAGCGGGACAGAGAAA-3’) and mc1r_R4 (5’-GAAGAGGAAGAAGCTGATGAG-3’), followed by nested PCR using the same forward primer (mc1r_F2) and a different reverse primer nested just 5’ to mc1r_R4 (mc1r_R3 5’-GATGGCATTGTTGCGATAGTA-3’). The 3’ region of Mc1r containing the stop codon was amplified using the primers mc1r_F6 (5’-CACCTGCAGCTCTGTTGTGT-3’) and mc1r_R6 (5’-ATGCCATTATCGGTGTCCCAC-3’). PCR products were gel extracted (Qiaex II Gel Extraction kit, Qiagen, Valencia, CA) and Sanger sequenced by the University of Washington High Throughput Genomics Unit using the primers MSHR9, MSHR72, mc1r_F2, mc1r_R4, mc1r_F6, and mc1r_R6. Sequences were assembled and analyzed using Sequencher software (Genecodes, Ann Arbor, MI). Of the 120 original birds in the panel, 113 from 67 breeds yielded usable sequences (94.2% success rate).

TaqMan Genotyping Assay

We genotyped a G253A mutation (coding for a Val85Met polymorphism) on 337 pigeons, including 77 birds used in the Sanger-sequenced SNP discovery sample, using a TaqMan assay with the following oligonucleotides: forward primer, 5’- CATCTGCTGCCTAGCCATCT C-3’; reverse primer, 5’-GCTCCATCAGCAGCATGAAGA-3’; probes 5’ VIC-ATGCTGGTGAGCGTC-3’ (Val allele) and 5’ FAM-CATGCTGATGAGCGTC-3’ (Met allele.) TaqMan assays were performed by the University of Utah DNA Genomics Core Facility using the manufacturer’s protocol (Applied Biosystems, Foster City, CA). We discarded samples with genotype calls that fell outside the 95 percent confidence interval, leaving a sample of 313 birds (92.9% success rate).

Association Tests

To reduce the possibility of ambiguous phenotypes based solely on descriptive information from breeders, we further filtered the dataset to exclude birds for which we did not also have photographs. This reduced the sample to 237 birds. We performed a final filter on the data to avoid pseudo-replicates in the dataset: we included only one bird per breeder of each breed with the same color and genotype, leaving a total of 190 birds from 89 breeds. Pigeons were scored for pigmentation phenotypes based on two categorical schemes. First, birds were classified as either eumelanic or pheomelanic based on their plumage color. A bird was classified as eumelanic if its major plumage coloration was a shade of brown or black, including gray. Pheomelanic birds were those with red or yellow plumage. Second, birds were categorized as blue/black, ash-red, brown (color phenotypes at the major sex-linked pigmentation locus); recessive red (autosomal, recessive alleles, and epistatic to the major color locus); and white (most common locus is autosomal with recessive alleles and epistatic to the major locus, but other autosomal loci can contribute as well) [23]. This more specific classification allowed us to test for enrichment of mutant alleles in blue/black, recessive red, and white birds.

We used GraphPad Prism software (GraphPad Software, San Diego, CA) to conduct tests for association between phenotypes and Mc1r genotypes using a chi-square test with 2 X 3 contingency tables. White birds were excluded from analyses except for tests of white versus all other colors. We excluded white birds because all-white pigmentation can be a complicated phenotype requiring the interaction of multiple loci [23].

Cloning of Val85 and Met85 alleles

We isolated and cloned Val85 and Met85 alleles of Mc1r to test the functional effects of the Val85Met substitution. We amplified alleles from genomic DNA of a black mottle West-of-England tumbler (Val85) and a white standard fantail (Met85) by PCR using primers containing HindIII and EcoRI restriction sites: Mc1r_F7_HindIII 5'-TCGATAAGCTTGTGCCCTGGAGCTGAGGT-3' (HindIII site in italics) and Mc1r_R7_EcoRI 5'-GTCACGAATTCGTGTCCCAC TGCCTACCAG-3' (EcoRI site in italics). Products of each of these reactions were then gel-purified (Qiaex II Gel Extraction kit, Qiagen) and cloned into the pcDNA3.1/V5-HisB vector (Invitrogen, Grand Island, NY). Allele sequences were confirmed to be identical at the nucleotide level except for the G253A SNP that codes for the variable Val85Met residue.

Functional testing of alternative alleles

COS-7 and HEK293T cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) and F12 medium, respectively, supplemented with 10% (v/v) fetal bovine serum, 100 units/ml penicillin and 100 µg/ml streptomycin, at 37°C in a humidified 5% CO₂ incubator. For expression in mammalian cell lines, the two Mc1r alleles were subcloned into the mammalian expression vector pcDps. To quantify protein expression and the amount of plasma membrane-integrated Mc1r, the two Mc1r alleles were double-tagged with an N-terminal HA (haemagglutinin) tag and a C-terminal FLAG-tag. All PCR-derived constructs were verified by sequencing. Transient transfection experiments were performed with Lipofectamine 2000 (Invitrogen).

For cAMP assays, transfected HEK293T cells were split into a 384-well plate (10,000 cells per well) and stimulated with αMSH. Concentrations ranging from 0.1 nM to 100 µM were administered in logarithmic increments. The cAMP content of each well was determined by a non-radioactive cAMP accumulation assay based on the ALPHAScreen technology according to the manufacturer’s protocol (PerkinElmer LAS, Rodgau-Jügesheim, Germany). A third construct containing GFP, which should exhibit no cAMP accumulation, was used as a negative control for the assay. Cell vitality was measured using a 10-µM forskolin treatment to ensure that the adenylyl cyclase of all cells was capable of producing cAMP. Cyclic AMP accumulation data were analyzed using GraphPad Prism.

To assess the expression of full-length HA/FLAG double-tagged Mc1r proteins, and to demonstrate that the levels of cell surface expression were not due to a decrease or increase in receptor expression in general, a sandwich ELISA was used and performed as previously described to measure total cellular expression [28]. Specific optical density (OD) readings (OD_{492/620 nm} value of HA/FLAG-tagged construct minus OD_{492/620 nm} value of GFP-transfected cells) are given as a percentage of the Val85 allele of Mc1r. Results are reported as means ± SD of three independent experiments, each performed in triplicate or quadruplicate.

Variation in Mc1r sequence

We identified 7 variants in the coding sequence of Mc1r in rock pigeons (Genbank accession numbers KF234242-KF234252). Of these coding variants, 3 were predicted to be synonymous and 4 non-synonymous at the amino acid level (Table 1). The most common haplotype was homozygous in 45.3% of birds, and heterozygous in 25.5%. When we further included haplotypes with only synonymous nucleotide differences, a total of 55.6% of birds were homozygous for the most common amino acid sequence, and another 36.8% were heterozygous. Four other haplotypes harbored a G253A nucleotide substitution, resulting in a Val85Met amino acid change (3.8% homozygous for A253, 7.5% heterozygous). When we examined variation within breeds, we noted that Met85 alleles were enriched in standard and Indian fantails (10% homozygous, 15% heterozygous), which are closely related to each other [27,29]. Both Val and Met are nonpolar and hydrophobic amino acids, but Met has a sulfur-containing side chain. The Val85 residue is highly conserved across vertebrates (Table 2), suggesting that it is probably important for normal Mc1r function [30].

Association of Val85Met with pigment variation

The Val85Met mutation in Mc1r is a particularly interesting candidate to influence pigment type in pigeons because it is strongly associated with eumelanic phenotypes in the lesser snow goose (Anser caerulescens caerulescens) and red-footed booby (Sula sula), either alone or in combination with another amino acid substitution on the same haplotype, respectively [6,7] (Figure 2). Therefore, we predicted that this mutation would be enriched in pigeons with eumelanic plumage pigmentation, and depleted in pigeons with pheomelanic pigmentation. To examine this hypothesis, we genotyped 190 pigeons from 89 breeds at codon 85 to test for associations with melanic phenotypes. First, we tested for an association between Mc1r genotypes and completely white plumage, but did not find one (χ² = 3.24, p = 0.20). Next, contrary to expectation, we did not find an enrichment of Met85 alleles in black or other eumelanic birds (Figure 3). However, we did find a significant enrichment of Met85 alleles in pheomelanic birds (eumelanic versus pheomelanic: χ² = 7.22, p < 0.03, n = 177, white birds excluded; black/blue – the phenotypes containing the highest proportion of eumelanin [31] – versus brown and pheomelanic colors: χ² = 3.27, p = 0.20; n = 177). We did not find an association between Mc1r genotypes and the pheomelanic recessive red phenotype (χ² = 3.11, p = 0.21, n = 177). We can therefore infer that Val85Met in Mc1r does not underlie the recessive red phenotype.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Amino acid sequence variation among avian orthologs of Mc1r.

Several variants have been implicated in pigment variation within avian species. The Val85Met mutation found in domestic pigeons is associated with eumelanism in the lesser snow goose and red-footed booby. Functional studies of Mc1r protein variants have been conducted for chicken [14] and pigeon (this study). Additional mutations in chicken have been identified on the E92K background [13] but are not shown here. Figure based on previous review of avian Mc1r diversity [2].

https://doi.org/10.1371/journal.pone.0074475.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Met85 alleles of Mc1r are associated with pheomelanism in the domestic rock pigeon.

A. Genotype frequencies across all sampled breeds, with birds categorized as eumelanic or pheomelanic. Met85 genotypes are significantly enriched in pheomelanic birds (p < 0.03, Chi-square test). B. Met85 genotypes are not significantly enriched in pheomelanic recessive red birds (p = 0.21, Chi-square test). C. Met85 genotypes are significantly enriched in pheomelanic standard and Indian fantails (p < 0.04, Chi-square test).

https://doi.org/10.1371/journal.pone.0074475.g003

We repeated the same association tests on a subset of our sample, the fantail breeds, in which Met85 alleles are found at a higher frequency than the overall sample. Here, too, we found an association between genotype at codon 85 and qualitative eumelanic versus pheomelanic phenotypic categories (χ² = 6.57, p < 0.04, n = 27). As with the overall sample, we did not find an association between genotypes and all-white plumage (χ² = 2.54, p = 0.28, n= 34; we did not have a sufficiently large sample of recessive red fantails to test for associations). When we removed the fantail sample from the species-wide sample, the association between Mc1r genotype and pheomelanic phenotype was not significant (χ² = 3.77, p = 0.15, n = 150).

In summary, we found an association between Met85 alleles and pheomelanic phenotypes in domestic rock pigeons, both across 89 breeds and in fantail breeds; however, the association in the species-wide sample depends upon inclusion of the fantails. These findings contrast with previous studies in other bird species, in which Val85Met was strongly associated with eumelanic phenotypes [6,7].

The Val85Met mutation reduces amount of Mc1r at the plasma membrane

Given the unexpected association between pheomelanism and Val85Met, we next tested the functionality of Val85 and Met85 alleles. Based on genetic associations with eumelanism in other birds, we would predict that the Met85 allele would have a higher functionality; however, based on the association with pheomelanism we found in pigeons, we would predict the opposite effect. To distinguish between these two possibilities, we transfected HEK293T cells with expression constructs coding for the most common Mc1r allele found in our variant discovery screen (Table 1), or an allele that differed only in the G253A SNP coding for the Val85Met polymorphism (no other synonymous or non-synonymous coding changes). We then assayed for production of cAMP in response to stimulation with the Mc1r agonist αMSH, and found that the Met85 allele produced significantly less cAMP relative to the Val85 allele (p < 0.02, paired t-test, n = 3 replicates; responses at highest agonist concentration: Val85 allele = 9.37 ± 1.41 pmol/mg protein, Met85 allele = 3.87 ± 0.50 pmol/mg protein) (Figure 4A). Because a decrease in cAMP production is correlated with pheomelanin synthesis [32], our functional tests suggest that the Met85 allele should not be enriched in eumelanic pigeons; indeed, this is consistent with our genetic association results.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Functional differences between Val85 and Met85 alleles of Mc1r.

A. cAMP production in response to αMSH stimulation is reduced in cells transfected with the Met85 allele relative to the Val85 allele (p < 0.02, paired t-test). Error bars, ± SEM. B. Total cellular protein expression of HA/FLAG-tagged Mc1r is equivalent in COS-7 cells transfected with pigeon Val85 and Met85 alleles. The non-specific OD_{492/620 nm} value (GFP) was 0.008 ± 0.001 (0% set point) and the OD_{492/620 nm} value of the Val85 allele was 0.415 ± 0.070 (100% set point). C. Surface protein expression of the Met85 allele is reduced relative to Val85 allele (p < 0.007, one-sample t-test). Non-specific OD_{492/620 nm} value (GFP) was 0.053 ± 0.050 (0%) and the OD_{492/620 nm} value of the Val85 allele was 0.501 ± 0.049 (100%).

https://doi.org/10.1371/journal.pone.0074475.g004

To further elucidate the mechanism underlying the functional discrepancy between alleles, we examined the cellular localization of Mc1r proteins containing either Val85 or Met85. To do this, we transfected COS-7 cells with HA/FLAG-tagged versions of the two alternative Mc1r alleles, and quantified expression by sandwich ELISA. We found that both alleles were expressed at similar levels overall (Met85 allele expression = 100.7% of Val85 allele, SD = 15.3; Figure 4B), but cell surface expression of the Met85 allele was only 57.3% of the Val85 allele (p < 0.007, SD = 5.9, μ_O = 100, one-sample t-test) (Figure 4C). Together, these experiments suggest that the Val85Met mutation results in reduced Mc1r-mediated production of cAMP, and this reduced functionality is probably due, at least in part, to reduced localization of Mc1r proteins to the plasma membrane.