Report

A Single Amino Acid Mutation Contributes to Adaptive Beach Mouse Color Pattern

See allHide authors and affiliations

Science  07 Jul 2006:
Vol. 313, Issue 5783, pp. 101-104
DOI: 10.1126/science.1126121

Abstract

Natural populations of beach mice exhibit a characteristic color pattern, relative to their mainland conspecifics, driven by natural selection for crypsis. We identified a derived, charge-changing amino acid mutation in the melanocortin-1 receptor (Mc1r) in beach mice, which decreases receptor function. In genetic crosses, allelic variation at Mc1r explains 9.8% to 36.4% of the variation in seven pigmentation traits determining color pattern. The derived Mc1r allele is present in Florida's Gulf Coast beach mice but not in Atlantic coast mice with similar light coloration, suggesting that different molecular mechanisms are responsible for convergent phenotypic evolution. Here, we link a single mutation in the coding region of a pigmentation gene to adaptive quantitative variation in the wild.

The identification of the specific molecular changes underlying adaptive variation in quantitative traits in wild populations is of prime interest (1, 2). Pigmentation phenotypes are particularly amenable to genetic dissection because of their high heritability and our knowledge of the underlying developmental pathway (3). In a series of classic natural history studies (4, 5), Sumner documented pigment variation in Peromyscus polionotus, including eight extremely light-colored “beach mouse” subspecies, which inhabit the primary dunes and barrier islands of Florida's Gulf and Atlantic coasts (6). This light color pattern is driven by selection for camouflage (7, 8) as major predators of P. polionotus include visual hunters (9). Because the barrier islands on the Gulf Coast are <6000 years old (10), this adaptive color variation may have evolved rapidly.

We examined the contribution of the melanocortin-1 receptor gene (Mc1r) to this adaptive color patterning. MC1R, a G protein–coupled receptor, plays a key role in melanogenesis by switching between the production of dark eumelanin and light pheomelanin (11). Mutations in Mc1r have been statistically associated with Mendelian color polymorphisms in several mammalian species (e.g., 1214) and in natural variants of avian plumage (15, 16).

We sequenced the entire coding region of Mc1r [954 base pairs (bp)] in five Santa Rosa Island beach mice (P. p. leucocephalus) and five mainland mice (P. p. subgriseus) from colonies derived from wild populations (17). A single, fixed nucleotide polymorphism (SNP) occurs between the two subspecies, resulting in a charge-changing amino acid variant (R65C) in the first intracellular loop of MC1R (fig. S1). To determine the ancestry of this mutation, we genotyped this SNP in 14 other Peromyscus species (N = 45 individuals) (table S1). These additional species are all fixed for R65 as in the mainland “dark allele,” suggesting that the “light allele” is derived and not present in any other fully pigmented Peromyscus species.

To examine independently the relationship between this mutation (R65C) and the derived light-colored phenotype, we generated a large reciprocal F2 intercross between the Santa Rosa Island beach mouse and the mainland subspecies (Fig. 1). We characterized color phenotype for seven pigmentation traits, which provide an overall description of the continuous variation in color pattern, and genotyped the Mc1r allele in 459 F2 individuals [126 with two dark Mc1r alleles (RR), 215 with one dark and one light allele (RC), and 118 with two light alleles (CC)]. Based on the observed phenotypic variation among F2s, beach mouse color pattern has a multigenic architecture (Table 1). Pairwise correlation (R2) between traits ranged from 0.147 to 0.500 (P < 0.05 for all comparisons), revealing a shared genetic basis among traits but also indicating the role of loci expressed in distinct spatial regions.

Fig. 1.

Photographs of a typical (A) mainland mouse (Peromyscus polionotus subgriseus) and (B) Santa Rosa Island beach mouse (P. p. leucocephalus), highlighting the color pattern differences in the flank, face, and rump between these subspecies.

Table 1.

Statistical association between allelic variation at Mc1r and seven pigmentation traits in 459 F2 individuals. Phenotypes were categorically scored based on the pigmentation pattern on individual hairs in seven areas (traits). For all seven traits, each F2 was scored as follows: 0, unpigmented hair; 1, partially pigmented hair; and 2, fully pigmented hair. The total number of F2s with each phenotypic score is partitioned by Mc1r genotype for each trait: RR, homozygous for the dark Mc1r allele; RC, heterozygous; and CC, homozygous for the light Mc1r allele. The frequency of F2s with each phenotype score is shown for each trait, with the phenotype of all F1 individuals in bold. The mean phenotypic score (±SD) for each trait is provided and is also partitioned among each genotype. The percentage of variance explained by Mc1r genotype (PVE) and likelihood ratio chi-square (X2) values are shown. *All seven tests were statistically significant (P < 0.0001).

Trait Mc1r genotype Phenotypic score Mean ± SD PVE (%) X2
0 1 2
Whisker 0.37 0.30 0.33 0.97 ± 0.84 11.5 115.3*
RR 12 49 65 1.42
RC 69 77 69 1.00
CC 85 13 20 0.45
Rostrum 0.11 0.18 0.71 1.61 ± 0.68 36.4 131.4*
RR 0 0 126 2.00
RC 0 40 175 1.81
CC 49 42 27 0.81
Cheek 0.14 0.60 0.26 1.12 ± 0.63 27.2 232.9*
RR 0 91 35 1.28
RC 0 159 55 1.25
CC 66 23 29 0.69
Eyebrow 0.16 0.25 0.59 1.42 ± 0.76 16.2 141.4*
RR 0 21 105 1.83
RC 18 62 135 1.54
CC 56 31 31 0.79
Ear 0.44 0.36 0.20 0.75 ± 0.77 10.2 98.0*
RR 28 53 45 1.13
RC 89 79 47 0.80
CC 85 33 0 0.28
Ventrum 0.69 0.11 0.20 0.51 ± 0.81 16.3 122.6*
RR 55 27 44 0.91
RC 144 24 47 0.55
CC 118 0 0 0.00
Ankle 0.45 0.35 0.20 0.75 ± 0.77 9.8 93.4*
RR 24 70 32 1.06
RC 94 66 55 0.82
CC 88 26 4 0.29

Two lines of evidence suggest that both dominant and recessive alleles contribute to the adaptive light color phenotype. First, all F1 hybrids are intermediate in overall color pattern, with some traits resembling the mainland parent and other traits the beach mouse parent (Table 1). Second, dominance varies at the Mc1r locus itself. Phenotypic scores for pigmentation traits among F2s are sometimes above the mean score of 1, suggesting that the light Mc1r allele is partially recessive, and sometimes below the mean score, suggesting that the light Mc1r allele is partially dominant (Table 1).

We found a significant statistical association between the R65C polymorphism and all seven traits, although the percentage of phenotypic variance explained by Mc1r varied (9.8% to 36.4%) (Table 1). For some pigmentation traits, the association between phenotype and Mc1r genotype was notable: All RR F2s (N = 126) expressed the darkest phenotype on the rostrum and, conversely, only CC individuals expressed the lightest rostrum phenotype (N = 49). We calculated principal component scores for the combined color traits and evaluated whether there were significant differences between the Mc1r alleles for PC1 scores (PC1 explained 71% of the variance in color traits); these analyses suggest that Mc1r accounts for 26% of PC1 (P < 0.0001). Together, these data indicate that Mc1r is a major effect locus on color pattern, having pleiotropic effects on pigmentation in spatially diverse areas of the body.

Functional tests of the light and dark Mc1r alleles demonstrate that the R65C amino acid mutation contributes to variation in adaptive pigmentation. We performed in vitro assays on HEK293T cells expressing light or dark Peromyscus coding region alleles at similar levels. Stimulation with an MC1R agonist (NDP-αMSH) and subsequent measurement of generated cyclic adenosine monophosphate (cAMP) by radioimmunoassay revealed that compared with dark receptor–expressing cells, cells that expressed the light MC1R had a statistically significant reduction in basal and stimulated cAMP formation (Fig. 2A) (P < 0.05, Student's t test, n = 4; maximum responses: dark allele = 120 ± 15 pmol/mg protein, light allele = 28 ± 5 pmol/mg protein). MC1R-mediated cAMP formation is associated with eumelanin production (18); thus, reduced cAMP production may underlie the lack of dark pigmentation observed in mice expressing the light MC1R.

Fig. 2.

Cyclic AMP and ligand-binding assays showing functional differences between light and dark Mc1r alleles. All data points are means ± SEM. (A) HEK293T cells transfected with dark MC1R (filled circles) or light MC1R (open circles) have different cAMP responses to NDP-αMSH stimulation. Bars indicate basal level of cAMP production in cells transfected with dark MC1R (black) and light MC1R (white); lines indicate standard error. (B) αMSH competition for [125I]NDP-αMSH binding to HEK293T cells transfected with dark MC1R (filled symbols) or light MC1R (open symbols). Squares indicate assays without GppNHp and circles with GppNHp (100 μM).

The light MC1R displays lower affinity binding to αMSH, and the decreased response to guanine nucleotide by the light receptor implies an altered ability to interact with G proteins (Fig. 2B). Specifically, whereas the dark receptor displays a substantial GppNHp-promoted shift in IC50 (from 1.37 ± 0.01 nM in the absence to 13.5 ± 0.1 nM in the presence of GppNHp), the light allele has a decreased affinity for nucleotide and a nonsignificant IC50 shift (without GppNHp, 222 ± 30 nM; with, 178 ± 14 nM). These results demonstrate that a single amino acid mutation in MC1R is responsible for reduced ligand binding and G protein coupling, consistent with the reduced ability of the light MC1Rs to promote cAMP formation in vivo.

We next examined the frequency of Mc1r alleles in natural populations by genotyping the informative SNP in eight beach mouse subspecies and one mainland subspecies (Fig. 3 and table S2). In the Gulf Coast the light allele was at highest frequency (0.95) in the palest beach mouse subspecies, absent in the darkest subspecies, and at intermediate frequencies in the other subspecies (0.05 to 0.85) (Fig. 3). The distribution of Mc1r allele frequencies among Gulf Coast beach mice is not correlated with geographic distance. The light Mc1r allele was not detected in the mainland subspecies (0 of 40 alleles). In natural populations, different combinations of functionally distinct Mc1r alleles may contribute to variation in color patterning between mainland and Gulf Coast beach mouse subspecies, among the Gulf Coast subspecies, and within subspecies.

Fig. 3.

Frequency of Mc1r alleles in one mainland and eight beach mouse subspecies from northern Florida. Images represent typical color patterns for each subspecies. Red areas represent the distribution of each beach mouse subspecies, and the gray area represents the range of mainland subspecies in Florida. Lines indicate sampling locales. Circles represent frequencies of the light Mc1r (white) and dark Mc1r (black) alleles identified. The number of alleles sampled is provided.

The derived Mc1r allele was absent from the similarly light-colored Atlantic coast beach mouse populations. We surveyed Mc1r allelic variation in two extant subspecies of beach mice and the extinct pallid beach mouse subspecies (∼300 miles east of the nearest Gulf Coast beach mouse population) (Fig. 3). The absence of the derived Mc1r SNP from all three Atlantic coast subspecies suggests that the same mutation does not contribute to convergent phenotypic adaptation and that light coloration has evolved independently on the Atlantic coast.

This work has specific implications for understanding the evolutionary mechanisms responsible for adaptive phenotypic change. First, the identification and functional characterization of a single amino acid mutation's effect on quantitative variation provides a convincing exception to a growing number of examples demonstrating that variation in morphology is governed by changes in gene regulatory regions (19, 20). Second, the observation that different combinations of alleles can produce similar pigmentation patterns suggests that distinct molecular mechanisms can underlie adaptive convergence even in similar selective environments (but see 21). Finally, Mc1r represents a large effect locus, containing a quantitative trait nucleotide (QTN) contributing to variation in fitness, consistent with the view that adaptation may often proceed by large steps.

Supporting Online Material

www.sciencemag.org/cgi/content/full/313/5783/101/DC1

Materials and Methods

Fig. S1

Tables S1 and S2

References

References and Notes

View Abstract

Navigate This Article