Eichler and colleagues used array-based analysis of DNA samples from about 2,500 healthy people to estimate copy number variation in the human population as a whole. This study stands out for its large sample size, which enables it to give an overview of this type of genetic idiosyncrasy in the population at large. First author Andy Itsara and colleagues report that most individuals (extrapolated to 65 to 85 percent of the general population) harbor a CNV that is at least 100 kb of DNA long. Much larger (>500 kb) variants occur in 5 to 10 percent of individuals, while at least 1 percent of the population carries a CNV exceeding 1 Mb. CNVs longer than 100 kb are rare, while those topping 500 kb tend to be found in only one individual. These findings are in keeping with the idea that huge CNVs are bad for your health. Counted from the other side, any given CNV is present in the population at a frequency of 0.2 to 1.0 percent.

Previous analysis of a sample set designed to measure human genome diversity on a global scale (see Cann et al., 2002) suggested that certain world populations carry more than their fair share of CNVs—20 to 30 per person compared to the average of seven to nine (see Jakobsson et al., 2008). Itsara and colleagues examined the same sample set, and while they confirmed that two of those three populations, Melanesian and Papuan, did have a higher prevalence (11.9, and 10.3 CNV per individual) than other populations, the difference was marginal. The third group, the Kalash, had fewer CNVs than average. “Deeper population screens to assess the distribution of large and rare CNVs in the human population are clearly warranted, because although such variants may segregate within specific populations because of genetic drift, others may contribute disproportionately to disease susceptibility or alternatively be adaptive within those populations,” write the authors.

CNVs are clearly linked to disease. Triplication of the entire chromosome 21, for example, gives rise to Down syndrome, which is accompanied by a much greater risk for dementia. Duplication of the amyloid precursor protein gene on chromosome 21 leads to early-onset, familial AD (see ARF related news story), while duplication or triplication of the α-synuclein gene causes familial Parkinson disease (see ARF related news story and Ibanez et al., 2009). Abnormally high CNVs have also been linked to schizophrenia and autism (see related story on Schizophrenia Research Forum).

To assess the impact of CNVs on some neurologic diseases, Itsara and colleagues combined their data with those from nine genomewide association studies of schizophrenia, autism, and mental retardation, assembling CNV data on 6,860 affected individuals and 5,674 controls. Their analysis recovered known associations (e.g., deletions at chromosome 22q11 in some schizophrenia patients) and also revealed new, unexpected relationships.

The Seattle geneticists found that a chromosome 17p11.2 microdeletion normally associated with a disease called heredity neuropathy with liability to pressure palsies (HNPP) is also deleted in patients with schizophrenia and autism. At a locus on chromosome 16p12 that is predicted to be a risk candidate for schizophrenia (see Stone et al., 2008), Itsara and colleagues found a deletion in one autistic patient and no deletions in any controls, again suggesting some overlap between autism and schizophrenia risk factors. And their data suggest that at chromosome 3q29, where a microdeletion leads to a syndrome that includes mental retardation and other neurologic abnormalities (see Willatt et al., 2005), deletions again increase the risk for schizophrenia. The findings tie clinically separate disorders together through the same CNV, leading the authors to suggest that these loci render their carriers generally vulnerable to mental illness such that the specific manifestation in a particular person depends on genetic modifier or environmental effects.

Exactly such genetic modification is what Barak Cohen and colleagues tried to come to grips with in their yeast study, which focused on a widely variable trait—sporulation. In Saccharomyces cerevisiae sporulation is a complex, heritable trait. It is subject to environmental influence, and believed to fall under different selection pressure in different environments, such as the oak grove and the oak barrel. Yeast from the former sporulate at near 100 percent efficiency, but those from the latter, perhaps not surprisingly, are much less competent. Sporulation serves as a model for complex traits, including susceptibility to disease and resistance to pharmacological intervention, exhibited by humans. Many individual genetic risk factors for late-onset Alzheimer disease have been identified, for example (see AlzRisk database ), but it is not clear how any of them interact with each other to increase or even decrease susceptibility.

First author Justin Gerke and colleagues identified four nucleotide changes among three transcription factors that explain the natural variation in yeast sporulation efficiency. The researchers crossed two parent strains, one from the North American oak and one from a California wine barrel, and looked for quantitative trait loci in the offspring that match their sporulation efficiency.

The researchers identified five loci that accounted for most of the variation among the different offspring. Three of these loci, all of which turned out to harbor transcription factors, had large effects. By sequencing the different strains, the researchers found four nucleotide substitutions that account for almost 80 percent of the sporulation variation: a single nucleotide change in the regulatory region of RME1, a transcription factor that can suppress sporulation in certain cell types; two non-synonymous substitutions in the coding region of IME1, a master regulator that initiates sporulation; and a single coding change in RSF1, transcriptional activator of mitochondrial genes essential for respiration. By replacing nucleotides one, two, three, and four at a time, Gerke and colleagues found that all four alleles interact to alter the phenotype.

“Knowing how individual genetic polymorphisms combine to produce phenotypic change could strengthen evolutionary theory and advance applications such as personalized medicine,” write the authors. In general, epistatic interactions between genes are poorly understood and this study highlights the effect that even single nucleotides can have on a given trait. “This emphasizes the need to incorporate genetic interactions into models that seek to accurately predict phenotype from genotype,” write the authors, adding that “if prevalent, genetic interactions between nucleotides will be a major hurdle in the endeavor to connect genetic and phenotypic variation in humans.”—Tom Fagan.

References:
Itsara A, Cooper GM, Baker C, Girirajan S, Li J, Absher D, Krauss RM, Myers RM, Ridker PM, Chasman DI, Mefford H, Ying P, Nickerson DA, Eichler EE. Population analysis of large copy number variants. Am. J. Hum. Genet. 2009, Jan 21. Abstract

Gerke J, Lorenz K, Cohen B. Genetic interactions between transcription factors cause natural variation in yeast. Science. 2009, Jan. 23; 323:498-501. Abstract