Ecological adaptation determines functional mammalian olfactory subgenomes

Abstract

The ability to smell is governed by the largest gene family in mammalian genomes, the olfactory receptor (OR) genes. Although these genes are well annotated in the finished human and mouse genomes, we still do not understand which receptors bind specific odorants or how they fully function. Previous comparative studies have been taxonomically limited and mostly focused on the percentage of OR pseudogenes within species. No study has investigated the adaptive changes of functional OR gene families across phylogenetically and ecologically diverse mammals. To determine the extent to which OR gene repertoires have been influenced by habitat, sensory specialization, and other ecological traits, to better understand the functional importance of specific OR gene families and thus the odorants they bind, we compared the functional OR gene repertoires from 50 mammalian genomes. We amplified more than 2000 OR genes in aquatic, semi-aquatic, and flying mammals and coupled these data with 48,000 OR genes from mostly terrestrial mammals, extracted from genomic projects. Phylogenomic, Bayesian assignment, and principle component analyses partitioned species by ecotype (aquatic, semi-aquatic, terrestrial, flying) rather than phylogenetic relatedness, and identified OR families important for each habitat. Functional OR gene repertoires were reduced independently in the multiple origins of aquatic mammals and were significantly divergent in bats. We reject recent neutralist views of olfactory subgenome evolution and correlate specific OR gene families with physiological requirements, a preliminary step toward unraveling the relationship between specific odors and respective OR gene families.

Figure 1.

Phylogenomic partitioning of olfactory receptor genes. (A) Bayesian phylogeny of functional mammalian olfactory receptor (OR) genes available in Ensembl v52. OR gene family names are labeled on the termini of each subtree. MCMC posterior probabilities are given for each node. Polyphyletic families are clustered in subsequent analyses. (B) Ratio of functional OR genes per family (number of functional genes per family/total number of functional genes per species). The two C. familiaris (dog) genome assemblies are not significantly different (χ² = 4.8, degrees of freedom [df] = 10, P ≈ 0.91). (C) Ratio of OR pseudogenes per family (number of pseudogenes per family/total number of pseudogenes per species). The two C. familiaris genome assemblies, boxer (7.6×) and poodle (1.5×), are significantly different (χ² = 79.0, df = 8, P < 0.0001). (D) Heatmap displaying the relative percentage of functional genes in each OR gene family of different mammalian genomes, mapped onto the consensus phylogenetic tree for mammals. Selected ancestral reconstructions are shown at the bottom.

Figure 2.

Ecotype partitioning by olfactory receptor genes. (A) Heatmap of the informative OR gene families partitioned by ecological group (color bars on right follow legend). (B) Naïve Bayes pattern analysis of OR gene repertoires showing species assignment into ecological groups: (pink) volant, (red) terrestrial, (blue) aquatic, or (green) semi-aquatic based on the familial distribution of OR genes. Species abbreviations follow A (color bars on top follow legend). (C) Scatterplots showing the results of PCA analysis on functional OR genes within their respective OR gene families. The first and second axes explain 70% of the variance within the data set (ANOSIM: r = 0.6095, P < 0.001). Colored polygons highlight the different ecological groups of mammals. (Green lines) The contribution of particular families to the first two principal components. Species abbreviations, including ancestors, follow A. (D) Close-up view of the terrestrial polygon in C (boxed).

Figure 3.

Naïve Bayes assignment test of OR gene repertoires showing species assignment into phylogenetic groups based on the familial distribution of OR genes. (Pink) Marsupialia/Monotremata, (red) Euarchontoglires, (green) Afrotheria, (blue) Laurasiatheria.

Figure 4.

Ecological specialization influences OR gene repertoires. Illustrated scatter plot showing the global percentage of pseudogenes within Cetartiodactyla, Carnivora, and Sirenia. The median values of the percentage pseudogenes from high coverage genomic and laboratory generated data for terrestrial, aquatic, and semi-aquatic taxa are depicted along the y-axis. The genomic pseudogenes are considered at 650 bp, which is similar to the length of the PCR product that we amplify. Low-coverage assemblies were excluded.