What is a functional genetic polymorphism? Defining classes of functionality

Abstract

In the human genome there are at least 3.1 million single nucleotide polymorphisms (SNPs), or about 1 SNP per kilobase of sequence.1 There have been thousands of studies in mental health, including genome-wide association studies, attempting to associate mental illness or various behavioural, imaging or biochemical endophenotypes with genetic polymorphisms or sequence variants (following recommendations of the Human Genome Variation Society, “sequence variant” is a more inclusive term than “polymorphism.” However, a common problem has been a lack of consistent association with a given polymorphism. For example, 5-HTTLPR is the most studied functional polymorphism in psychiatry, yet controversy remains as to the strength of association, with both positive and negative findings having been reported.2 In psychiatry in particular, there remains a mismatch between association of SNPs with disease and their functional role in disease. To begin to link SNP associations with potential functional roles, I propose a classification system that ranks the extent to which a given SNP has been demonstrated to have a functional role, with the highest rank being its role in the behaviours that result in mental illness. However, the proposed classification does not rank the magnitude of functional impact of a polymorphism, and hence cannot predict the likelihood of association of a given SNP with mental illness. To understand the mechanistic basis by which a polymorphism is associated with a particular phenotype or behavioural outcome, it is necessary to know whether that polymorphism is functional (i.e., whether it alters the function of a gene or set of genes). In most cases, the function of an associated polymorphism is not defined and must be surmised or extrapolated as an effect on the gene that contains this polymorphism. In rare cases, a polymorphism may be a nonsynonymous coding region variation that alters the gene product protein structure. Most common polymorphisms are potential regulatory polymorphisms located in noncoding regions, including promoter/upstream, downstream and intron regions, that may affect transcription;3 in intron and untranslated regions transcribed as RNA that may affect transcription, RNA splicing, stability or translation;4 or in intergenic regions of unknown function.5 For example, 2 SNPs in the dopamine D2 receptor gene introns 5–6 alter its splicing to favour the generation of the D2 long over short receptor isoform and are associated with reductions in working memory and reduced frontostriatal activation in people with schizophrenia and people who abuse cocaine.6–8 Individual SNPs may have minimal functional impact but may be in linkage disequilibrium with a set of polymorphisms that form a haplotype associated with a functional outcome on gene expression or function.9 Whereas a polymorphism may have a demonstrated effect on the function of a gene, there are varying degrees of demonstrating the function of a genetic variant. These range from in vitro studies to studies that seek to determine the functional impact of a given polymorphism on gene expression in humans as the clearest indication of a “functional polymorphism.” Hence, for the purposes of generating discussion on this issue, a potential classification system to rank the function of a polymorphism or sequence variant is presented. I propose 4 classes of genetic polymorphisms: Class 0: Function not determined. Either (A) no function is known, or (B) theoretical function is predicted but has not been experimentally demonstrated. Class 1: Functional in vitro. The functional effect of the polymorphism on a target DNA element or regulatory mechanism has been demonstrated using in vitro assays (e.g., gel shift, reporter assay, ligand binding); however, the function of the polymorphism on endogenous gene expression or in vivo is unknown. Class 2: Functional in vivo. In addition to class 1 requirements, (A) function effect of the polymorphism on the endogenous gene has been tested in model cellular systems (e.g., human transformed cell lines, human B lymph-oblasts, primary cell cultures) using methods such as relative allelic expression and chromatin immunoprecipitation, and (B) in vitro function is correlated with a functional change in human tissue. Class 3: Functional phenocopy. In addition to class 1 requirements, (A) function has been demonstrated in vivo using model organisms such as knockin mice, and (B) function is correlated with a functional change in human tissue. Note that within each class, the degree of functional impact may vary. Hence, the class ranking does not indicate the magnitude of functional impact of the polymorphism; rather, it indicates only the degree to which the impact has been investigated. Furthermore, the class of a polymorphism does not imply its importance in a given population: a polymorphism may have a large impact on function in vivo, but be exceedingly rare and not useful as a marker in the general population, as is the case for TPH2 G1463A.10,11