Extensive copy-number variation of the human olfactory receptor gene family

Abstract

As much as a quarter of the human genome has been reported to vary in copy number between individuals, including regions containing about half of the members of the olfactory receptor (OR) gene family. We have undertaken a detailed study of copy-number variation of ORs to elucidate the selective and mechanistic forces acting on this gene family and the true impact of copy-number variation on human OR repertoires. We argue that the properties of copy-number variants (CNVs) and other sets of large genomic regions violate the assumptions of statistical methods that are commonly used in the assessment of gene enrichment. Using more appropriate methods, we provide evidence that OR enrichment in CNVs is not due to positive selection but is because of OR preponderance in segmentally duplicated regions, which are known to be frequently copy-number variable, and because purifying selection against CNVs is lower in OR-containing regions than in regions containing essential genes. We also combine multiplex ligation-dependent probe amplification (MLPA) and PCR to assay the copy numbers of 37 candidate CNV ORs in a panel of approximately 50 human individuals. We confirm copy-number variation of 18 ORs but find no variation in this human-diversity panel for 16 other ORs, highlighting the caveat that reported intervals often overrepresent true CNVs. The copy-number variation we describe is likely to underpin significant variation in olfactory abilities among human individuals. Finally, we show that both homology-based and homology-independent processes have played a recent role in remodeling the OR family.

Figure 1

Copy-Number Assessment of Three Representative ORs by MLPA Normalized peak heights (means of triplicate measurements) are shown for three MLPA probe pairs, each surveyed in 46 individuals. Peak heights are proportional to genomic copy number. For each probe pair, samples are sorted in ascending order of peak height. Numbers in gray represent mean peak heights for each group of samples with the same allelic state, with standard deviations for each group in parentheses and inferred copy number given as, for example, CN2 (copy number = 2). (A) The OR8U8 probe pair reveals a polymorphic deletion. (B) The OR4K2 probe pair reveals a polymorphic duplication. Note that the ratio of the mean peak heights of the genotype groups we have assigned is 2: 3.0: 4.0: 5.9. We treat the copy-number assignment of the individual who appears to have 6 copies tentatively at present, as well as two other allelic states only observed in a single individual in our panel (three copies of OR2G6, three copies of OR2T11). Sampling a much larger panel of individuals, and/or DNA extracted independently from the individual in question (not derived from an immortalized cell line) would be needed to increase confidence in these observations. (C) The OR6C1 probe pair does not reveal any copy-number variation, showing very consistent peak height across the panel of 46 individuals we surveyed.

Figure 2

Number of Functional Copies of 19 OR Genes Examined in a Panel of 51 Individuals We summarize in this figure both copy-number variation and single-nucleotide polymorphisms that would disrupt OR function (“segregating pseudogenes”), in some cases in the same gene. Figure S2 shows copy-number variation alone. Each row represents one of the human individuals tested as part of our diversity panel. Each column of the grid summarizes genotype data for an OR gene or, in three cases (^∗), for groups of OR genes (see below). Table 2 details the polymorphisms summarized in this figure. Genes are ordered according to the number of copies gained or lost, averaged over the individuals surveyed. Full genotype data are also given in Table S7. ^∗ “OR8U8,etc”: a deletion CNV destroys function of OR8U8 and OR8U9, while simultaneously creating a novel hybrid gene, OR8U1; “OR56B2, etc”: a deletion removes all of OR56B2 and OR52N5 as well as half of OR52N1; and “OR4C11, etc”: a complex set of deletions removes OR4C11, OR4P4, OR4S2, OR4V1P and OR4P1P (Figure S1).

Figure 3

Cartoons of Genomic Structures and Breakpoint-Sequence Alignments of Two Representative Sets of Alternative Structural Alleles (A) Nonallelic homologous recombination (NAHR) between duplicated ∼10.7 kbp sequence blocks with 84% identity, of which 180 bp is shown aligned here. NAHR appears to have mediated a deletion that removes OR9G9. The “crossover” occurred somewhere within the 36 bp of identical sequence indicated as “NAHR region.” (B) Nonhomologous end joining (NHEJ) mediated a deletion of OR56B2, OR52N5, and half of OR52N1—two bases of microhomology are indicated at the deletion breakpoint. In both (A) and (B), derived sequences spanning the deletion (middle rows) are shown aligned with the two ancestral sequences (outer rows). White letters on a black background indicate identical sequence; black letters on a white background indicate mismatched bases; “-“ symbols indicate alignment gaps.