Outside the Fold

 

 

Of the 20,000 or so genes carried in each and every one of our cells, only a limited number are expressed – “turned on” to produce proteins – at any time in any given cell type. Scientists have long sought to define the basic rules of gene expression, which is involved in everything from embryonic development to the growth of cancer. But the rules are anything but basic: Expressing the right genes at the right time requires a host of attendant molecules to regulate the process – from the proteins that initiate the copying of information encoded in the genes, to the different molecules that act on that coded information at various stages along the way.

 

One of the later-stage regulatory molecules is microRNA. MicroRNAs are, as their name implies, short bits of RNA that latch on to messenger RNA – the single-stranded molecules that ferry the gene code out of the cell’s nucleus to its protein factories – and prevent them from producing proteins. Hundreds of different microRNAs have been identified to date, and each homes in on the expression of a different set of genes. A host of diseases, including some cancers, may involve errors in microRNA activity.

 

To get a complete picture of microRNA regulation, researchers should ideally be able to identify the microRNAs’ target sites along the length of the messenger RNA molecules. To this end, many scientists have searched for short sequences of messenger RNA that match up with complementary microRNA sequences, much as the two sides of the double-stranded DNA molecule fit together. Unfortunately, it turns out that microRNA sequences are generally not a perfect fit for their target sites, and methods for predicting which segments of messenger RNA contain target sites have not stood up to lab experiments.

 

Dr. Eran Segal and research student Michael Kertesz of the Computer Science and Applied Mathematics Department decided to tackle the problem from another angle altogether: Instead of focusing on the linear sequence, they looked at the three-dimensional structure of the messenger RNA molecule. Like many biological molecules, messenger RNA has a tendency to fold up like abstract origami. To find out whether this folding affects where microRNAs bind to the messenger RNA, Segal and Kertesz, together with a research group from Rockefeller University, New York, performed an experiment. They created a series of mutations in cells so that the messenger RNA either folded tightly around possible target sites or opened up to expose them. The researchers then compared the original levels of gene expression with those of genes in the mutated cells. “If the existing theory is correct, these mutations shouldn’t affect gene expression, as the sequences that are in direct contact with the microRNAs remain the same,” says Segal. But the results of their experiment clearly indicated that changing the shape of the folded messenger molecule does affect levels of gene expression. When the target site was open and accessible to the microRNA strand, binding activity rose and protein production came to a screeching halt. In contrast, closing off the target site with tight folds resulted in less binding between the molecules and, in consequence, rising levels of gene expression.

 

On the basis of these results, the scientists developed a mathematical formula for identifying target sites. This formula calculates the difference between the energy that must be invested in opening up a potential target site and the chemical energy released when the microRNA binds to the messenger RNA sequence. The more tightly the site is folded, the higher the investment that is needed, whereas a better fit between the two sequences will cause more energy to be released upon binding. When the resulting difference is large – that is, there is a relatively high net energy gain in the process – chances are the site is a target site.

 

The scientists also found that for the most efficient binding, the open segment should extend beyond the sequence of the target site. This is because the microRNA is embedded in a large protein structure, and the open segment must be long enough to encompass the bigger mass. When they adjusted the formula to fit the protein’s size, they found their model to be a great improvement over the old methods for predicting target sites. A test of their model on potential target sites in the genomes of humans, mice, flies and worms showed that the spatial structure of the messenger RNA is, indeed, a determining factor for the position of target sites. The results of the study, which appeared in Nature Genetics, did not take every possible factor affecting molecular interactions into consideration. Nonetheless, says Segal, it should give scientists a useful tool for identifying likely target sites for microRNA binding.

 

Kertesz: “The beauty of this research is that despite all of the elements we didn’t factor into the formula, our simple model turns out to be a quite a good fit for the experimental results.”