These findings, by researchers based in Boston, Philadelphia and Oxford, are published December 5 in the open-access journal PLoS Genetics.

Gene expression is known to vary among individuals and to be influenced by both genetic and environmental factors. Previous studies have reported gene expression differences among human populations, but it has been suggested that this could be due to non-genetic effects. Populations of recently mixed ancestry such as African Americans, who on average inherit about 80% African and 20% European ancestry, offer a solution to this question, since individuals vary in their proportion of European ancestry while the analysis of a single population minimizes non-genetic factors.

In this study, the researchers show that gene expression levels in African Americans vary as a function of each individual's proportion of European ancestry. The differences due to ancestry (i.e. population differences between all Africans and all Europeans) were generally small ”much smaller than differences between individuals within the same population; nevertheless, the authors were able to draw a distinction between effects of genetic ancestry at the location of the expressed gene (cis) and genetic ancestry elsewhere in the genome (trans). They conclude that only about 12% of heritable variation in human gene expression is due to cis regulation.

First author Alkes Price says, "It was a surprise that these conclusions could be drawn given that the differences due to genetic ancestry are so small." However, he cautioned that the results were confined to gene expression levels in a particular type of tissue known as lymphoblastoid cell lines, and have yet to be verified in other tissue types.

plos/

"Why do organisms have reactions they don't use? Presumably so they have the flexibility to adapt to different conditions. One optimization situation will engage a certain set of reactions, while another situation will require a different set. It remains to be demonstrated, however, whether different conditions alone can justify the presence of all available reactions. The fact that this question is yet to be answered makes the entire problem even more attractive," Motter said.

Motter and his team used computational results to re-interpret and explain specific recent experimental results. First they gathered extensive experimental information on the metabolic networks of four different single-celled organisms: three bacteria ( H. pylori , S. aureus and E. coli ) and yeast ( S. cerevisiae ). Then the researchers built general quantitative models of the organisms that allowed them to predict cellular behavior. With those models, the researchers conducted mathematical analyses and computer experiments, simulating the organism and its metabolic function under optimal and non-optimal conditions.

They observed that for all four organisms in a typical non-optimal state, all utilizable reactions in the metabolic network, with a few exceptions, were active. In contrast, when the four organisms were growing at their optimal rate, each of them spontaneously silenced a large number of metabolic reactions. The number of active reactions, around 300, was the same for all four, despite differences in the size and complexity of each organism's genome and metabolic network. And the number stayed around 300 for a variety of quite different optimization scenarios.

"Mathematical abstraction of the problem suggests that spontaneous shutdown may not be limited to metabolic networks," said Nishikawa, who led the mathematical part of the effort. "What appears to be essential for this phenomenon is that a complex network that is under constraints and locally in balance is 'trying' to optimize its function. There are other important systems, like transportation networks, where the same type of analyses could be useful."

northwestern/