Although people often struggle to master more than one discipline, our genes are accomplished polymaths. Genome-wide surveys of gene expression in 15 different tissues and cell lines have revealed that up to 94% of human genes generate more than one product.

The surveys, published online on 2 November in Nature1 and Nature Genetics2, used high-throughput sequencing to generate the most detailed portrait yet of how genes are expressed in different tissues.

Only about 6% of human genes are made from a single, linear piece of DNA. Most genes are made from sections of DNA found at different locations along a strand. The data encoded in these fragments are joined together into a functional messenger RNA (mRNA) molecule that can be used as a template to generate proteins.

But researchers have found that the same gene can be assembled in different ways, sometimes leaving out a piece, for example, or including a bit of the intervening DNA sequence.
More complex than a nematode

This process, called alternative splicing, can produce mRNA molecules and proteins with dramatically different functions, despite being formed from the same gene. The phenomenon provides some solace to those disappointed by the relatively small number of genes in the human genome: with around 20,000 genes, humans have roughly the same number as the elegant but decidedly less complex nematode, Caenorhabditis elegans.

"We were expecting that something as sophisticated, complex and intelligent as ourselves would have about a hundred thousand genes at least," says Jacek Majewski, a genomicist at McGill University in Montreal, Canada. "Then we sequenced the genome and realized it was about the same number as C. elegans." Fortunately, alternative splicing is thought to occur in only about a tenth of C. elegans genes, restoring the dignity of complexity to the human genome. Understanding this flexibility should help to reveal how improperly spliced genes can trigger disease.

Despite intense interest in alternative splicing, the phenomenon has been difficult to study, and the usual laboratory techniques often fail to detect rare splice forms. Researchers previously estimated that 74%3 of all human genes are alternatively spliced, but recognized that this estimate was likely to increase as techniques to study the process improved.

Now two groups, one led by computational biologist Christopher Burge of the Massachusetts Institute of Technology in Cambridge and the other led by molecular biologist Benjamin Blencowe of the University of Toronto in Canada, have studied alternative splicing using high-throughput sequencing data generated by Illumina, a biotechnology company based in San Diego, California.

The technique works by using an enzyme to convert mRNA back to DNA, which can then be sequenced. Blencowe and his colleagues studied splice forms found in six different tissues, including the brain, liver, muscle, and lungs. Burge and his colleagues used these samples along with several others, including breast cancer cell lines. Based on over 400 million sequences, Burge's team estimates that 92–94% of all human genes can yield more than one RNA molecule.
Technological challenge

Specialists in the field agree that the work is important, but are not particularly surprised by the numbers. "What is new is the technology, which will have a big effect on how we study splicing," says Douglas Black, a molecular biologist at the University of California, Los Angeles.

Analysis of the new splicing catalogues can reveal patterns about how the process is regulated, but more work is needed to determine whether all of these different splice forms have a function. "The question now is, 'Are all of those forms biologically relevant?'" says Marie-Laure Yaspo, a genomicist at the Max Planck Institute for Molecular Genetics in Berlin, Germany. A few of those rare splice variants may be no more than background noise generated by occasional mistakes, she notes.

But conventional techniques for deleting entire genes are not effective for sorting out the function of one splice variant versus another. "What really needs to be done is to develop high-throughput methods for analysing the function of these splice variants," says Blencowe. "That's the big challenge ahead."