Personal receptor repertoires: olfaction as a model

Abstract

Background: Information on nucleotide diversity along completely sequenced human genomes has increased tremendously over the last few years. This makes it possible to reassess the diversity status of distinct receptor proteins in different human individuals. To this end, we focused on the complete inventory of human olfactory receptor coding regions as a model for personal receptor repertoires.

Results: By performing data-mining from public and private sources we scored genetic variations in 413 intact OR loci, for which one or more individuals had an intact open reading frame. Using 1000 Genomes Project haplotypes, we identified a total of 4069 full-length polypeptide variants encoded by these OR loci, average of ~10 per locus, constituting a lower limit for the effective human OR repertoire. Each individual is found to harbor as many as 600 OR allelic variants, ~50% higher than the locus count. Because OR neuronal expression is allelically excluded, this has direct effect on smell perception diversity of the species. We further identified 244 OR segregating pseudogenes (SPGs), loci showing both intact and pseudogene forms in the population, twenty-six of which are annotatively "resurrected" from a pseudogene status in the reference genome. Using a custom SNP microarray we validated 150 SPGs in a cohort of 468 individuals, with every individual genome averaging 36 disrupted sequence variations, 15 in homozygote form. Finally, we generated a multi-source compendium of 63 OR loci harboring deletion Copy Number Variations (CNVs). Our combined data suggest that 271 of the 413 intact OR loci (66%) are affected by nonfunctional SNPs/indels and/or CNVs.

Conclusions: These results portray a case of unusually high genetic diversity, and suggest that individual humans have a highly personalized inventory of functional olfactory receptors, a conclusion that might apply to other receptor multigene families.

Figure 1

A summary of the genomic variation counts in intact OR coding regions and in OR pseudogenes.A, The absolute count. B, Count normalized per gene. Intact genes, blue and light blue; pseudogenes, orange and yellow. Nonfunctional variations are indicated by arrows. Abbreviations: miss, missense SNP; indel, small insertion/deletion up to 100 bp; del, CNV deletion; dup, CNV duplication; inv, CNV inversion; stop, stop gain/stop loss/ loss of the initiating methionine.

Figure 2

OR protein haplotype alleles for selected ORs. This is shown for OR1D2 (A), OR4E2 (B) and OR7C2 (C), typifying genes with high inter-allele diversity of CORP-predicted functionality. Segregating protein positions (indicated on top) are shown for each haplotype sequence, with yellow indicating non-reference SNP allele. The ancestral chimpanzee allele is shown in the lower row of each panel. The frequency of each allele in the population (%freq) and the CORP pseudogene probability score [43] are indicated in the two right columns. A high CORP score predicts a high pseudogene probability.

Figure 3

Protein allele genotype for 30 selected OR genes in 30 individuals. The ORs and individuals were selected to show maximal inter-allele diversity of CORP-predicted functionality. The two allelic protein sequences at each locus are shown, color-coded by their CORP scores for missense, and as indicated by the abbreviations (see Figure 1) for nonfunctional, and as depicted by the color scale on right, Ethnicities: 1–11 Europeans, 12–26 Africans, and 27–30 Asians.

Figure 4

Genomic variation for the entire OR intact haplotype repertoire in 145 individuals. Every individual is represented in every locus by a single randomly selected missense allele, except for Stop loci for which the non-reference allele is preferably shown. Color coding as in Figure 3. Ethnicities: 1–53 - Asians, 54–95 - Europeans and rows 96–145 - Africans. The dataset does not include alleles with concomitant indels and CNV deletions.

Figure 5

Population differences of personal OR protein allele counts.A) Distribution of the OR missense allele count frequencies in Africans (red), Europeans (brown) and Asians (blue). The black line indicates the average distribution for the whole population. B) Haplotype allele frequencies for six OR genes that show the highest inter-population variability. Only alleles with 1000 Genomes frequency > 10% in the entire human population are shown. AFR- Africans, ASN- Asians, EUR- Europeans.

Figure 6

A status diagram of the human OR repertoire. Among the 851 human OR sequences in the reference human genome, 464 were originally annotated as pseudogenes (P) and 387 as intact genes (I). Our study suggests that 218 (56%) of these intact ORs are segregating pseudogenes (S, originating from stop-SNPs and frame-disrupting indels), and 27 (6.9%) have a CNV deletion allele (D). Additionally, 26 (5.6%) of the OR pseudogenes are “resurrected (R)”, by showing an intact allele in some individuals.

Figure 7

Deletion CNV events in the human OR repertoire.A) The deletions size plotted against the deletion frequency in the 145 individuals analyzed. Circle size represents the number of OR genes affected by the deletion. B) Genotype calls for the 45 biallelic deletion loci [30] in 145 individuals. Black, homozygotes; grey, heterozygotes.

Figure 8

Personalized OR repertoires in 145 individuals. Blue- homozygotes for an intact allele, red- homozygotes for a disrupted allele, yellow- heterozygotes. Nonfunctional allele calls: stop SNPs, indels and deletion alleles.

Figure 9

Nonfunctional allele genotypes for 20 OR genes in 145 individuals. The genes have been selected to maximally span the genotype range. Individuals are sorted by ethnicity as in Figure 4. Allele statuses are: intact (I), nonfunctional SNPs/indels (S), bi-allelic deletion CNV (D) [30]. Colors indicate genotypic combinations. The full matrix with all 177 ORs in 145 individuals is shown in Additional file 1: Figure S7.

Figure 10

Population differences of OR SPGs.A) Principal component analysis of the nonfunctional SNP genotypes. Each point represents a specific individual, colors as in Figure 5A. B) Normalized relative frequencies of the nonfunctional OR allele in the three ethnic populations, color-coded as in (A). This is shown for 25 ORs, selected to represent the highest inter-population variability (values are given in Additional file 5). This include 20 ORs belonging to class II (“tetrapod-like”), members of 15 subfamilies (e.g. 1E), and 5 ORs belonging to class I (“fish-like”), represented by members of 5 subfamilies (e.g. 51F). OR classification is as described [3]. Colors as in Figure 5A.

Figure 11

OR genes are enriched with non-synonymous SNPs. Each panel compares the frequency distribution (f) for ORs (orange) and control genes (green). The analysis is done using three data sources: 1000 Genomes Project (A, B) with 581 genes with a single coding exon as controls; the GeneCards database [50] (C, D), with 15,425 protein coding genes as controls; all 10 data sources (Additional file 1: Table S1) (E, F), with OR pseudogenes as controls. A, C, E, non-synonymous SNPs, B, D, F, synonymous SNPs.

Figure 12

Selection signatures in the OR genes. Correlation of non-synonymous to synonymous substitution rate (pN/pS) with Tajima’s D values for A, 364 intact OR genes and B, 439 single coding exon genes. Data are plotted for the European population, other populations in Additional file 1: Figure S4. The difference between the ORs and the control genes was tested using the Kolmogorov-Smirnov test yielding p = 1.936*10^-24 for pN/pS, and p =2.26*10^-5 for Tajima’s D. The yellow squares highlight regions which might act under non-neutral selection, top right with D>1, pN/pS>1.5 (balancing selection), and bottom left with Tajima’s D<−0.5 and pN/pS<0.5 (purifying selection). Additional file 1: Table S3 lists the 57 ORs found under purifying selection in all populations.

Figure 13

A distribution of the pairwise inter-individual OR missense allele count differences for all 145 individuals shown in the previous figures.