Volume 21, Issue 1, March 2005

Table of Contents 21-1

Variation in Expression in Human Genes

• This Article
• print version (pdf)
• Printer Friendly
• Email Paper

Medical genetics textbooks typically distinguish between continuous and discontinuous variation in (clinical) phenotype. The latter can often be traced to a single change in the DNA sequence of a gene, ie, a “mutation” that serves as the basis of classic mendelian disorders. The genetic basis of continuously variable traits, such as height or blood pressure, is more difficult to explain. Variation in baseline expression of genes represents a mechanism that could contribute to continuous phenotypic variability. It is known to exhibit familial aggregation suggesting that it is heritable, but the tools to study the genetics of variation in human gene expression have only recently made it feasible to explore this notion. Morley and colleagues now document the existence of regulators of baseline gene expression.

The investigators utilized microarray technology to measure expression levels of genes, which they refer to as “gene expression phenotypes,” in immortalized B cells from members of 94 Center d’Etude du Polymorphism Humain (CEPH) Utah families. Starting with approximately 8500 genes active in these cells, they found 3554 genes that showed greater variation of expression between individuals than between replicates from the same individual. They then carried out genome-wide linkage analysis using single nucleotide polymorphisms to identify the genetic determinants of this variation. The results showed that variation in expression of 984 genes was genetically linked to one or more regions of the genome.

They assumed that regions linked to expression levels were regulatory regions or “regulators”. They examined the spatial relationship of the regulators to the 142 “target” genes that exhibited the strongest evidence for linkage. Twenty seven (19%) mapped to within 5 Mb of the target gene; they considered these to be cis- acting regulators because of their relatively close proximity to the coding sequence of the target gene. One hundred ten (77.5%) mapped further away and were designated trans-acting regulators. Both cis- and trans-acting regulators were found for 5 (3.5%) of the variably expressed genes. Many of these genes (164/984, or 16%) had multiple regulators of expression.

In addition to genomic regions containing regulators that influence single expression phenotypes in cis or in trans, the authors also found genomic regions that contained transcriptional regulators of multiple expression phenotypes. To further characterize these regulators, they divided the genome into 5 Mb windows and searched for regulatory “hotspots” within these windows. Two hotspots were detected, one of which mapped to chromosome 14 (14q32) and the other to chromosome 20 (20q13). Further analysis showed that these 2 regulatory hotspots influence expression of 31 of the 984 target genes under investigation. The authors suggest that their existence provides evidence for master regulators of baseline gene expression in humans.

Finally, they asked if differential expression of target gene alleles could be explained by cis-acting regulators. Analysis of individuals in whom alleles could be distinguished by single nucleotide polymorphisms showed that some of the variable expression could be attributed to the influence of the cis-acting regulators.

Morley M, Molony CM, Weber TM, et al. Genetic analysis of genome-wide variation in human gene expression. Nature 2004;430:743-7.

Cox NJ. An expression of interest. Nature 2004;430:733-4.

Editor’s Comment: This paper reminds us that the level of expression is an important aspect of gene action. Reduced or increased gene expression can influence quantitative traits, such as height. One can also envision a situation in which a mutation in a trans-acting regulator could cause disease by decreasing or increasing expression of its target gene(s). Take osteogenesis imperfecta type I for example; it typically results from mutations that cause transcripts from a mutant COL1A1 allele to terminate prematurely or undergo nonsense-mediated mRNA decay, functionally inactivating one of the 2 COL1A1 alleles. It is conceivable that a loss of function mutation of a trans-acting regulator of this locus could produce a similar adverse effect on type I collagen synthesis, especially if it were homozygous. Of note, such a mutation would not show linkage to the COL1A1 locus. There are several limitations of this investigation as noted by the authors and an accompanying news and views article. For instance, mRNA levels are only one determinant of the level of protein encoded by a given gene. Gene expression differs in different tissues, at different developmental stages and in response to physiologic and pathologic factors that are probably not reflected in immortalized B cells.

William A. Horton, MD

• This Article
• print version (pdf)
• Printer Friendly
• Email Paper