Two independent projects in different countries and with very different subjects came to the same conclusion at around the same time. Both pinpoint a gene as having a crucial role in the formation of the insulin-producing β-cells of the pancreas — the cells that are defective in some forms of diabetes.

Constantin Polychronakos (left) and Michael German.

Although there are various causes, diabetes ultimately results from insulin dysregulation. In one of the two most common forms of the disease, type 1 diabetes, the immune system attacks the pancreatic β-cells, progressively destroying them. Patients are treated with insulin injections, but scientists hope one day to be able to produce β-cells in the lab and then introduce these into patients to replace faulty or damaged cells.

Constantin Polychronakos, a geneticist at McGill University in Montreal, Canada, has a long-standing interest in the genes that predispose people to type 1 diabetes. But in 2004, during clinical rounds as a paediatric endocrinologist, he was intrigued by a rare form of neonatal diabetes in two siblings. As he described in a publication that year, each child's pancreas completely lacked insulin-producing β-cells, as though the body had not had the proper instructions for making them.

“Because the patients were siblings, had exactly the same disease and had parents who were second cousins, it seemed likely that the disease was caused by a mutation in a single gene inherited from both parents and derived from the same great-grandparent,” says Polychronakos. To begin the hunt for this mutation, Polychronakos and his team used a technique called homozygosity mapping. Every child inherits one half of each chromosome pair from each parent. Depending on how closely related the parents are, some regions in each chromosome pair will be identical, or homozygous. “In children whose parents are second cousins, 3% of the genome is homozygous,” says Polychronakos. He reasoned that the mutated gene inherited by the siblings with neonatal diabetes was almost certain to reside in one of those homozygous regions.

We now have a more complete picture of how to make a normal, mature β-cell.

Polychronakos and his colleagues went on to identify the regions of homozygosity in the two siblings' genomes, and in that of another child who had since been identified with the same disease in the United Kingdom. This allowed them to narrow down the list of candidate genes to about 200. “That is a heck of a lot of genes,” Polychronakos admits. “But with current sequencing technologies it was feasible to sequence them all to find the mutation.”

At around that time, in September 2007, Polychronakos received an e-mail out of the blue from Michael German, a diabetes researcher at the University of California, San Francisco. German's lab was examining the function of a mouse gene called Rfx6 that he had a hunch might be responsible for the unusual condition of the two siblings that Polychronakos had described.

German's lab was trying to define the mechanisms by which insulin-producing β-cells develop. One gene, neurogenin 3, was known to 'switch on' several other genes that then worked together to make a β-cell. Some of those had already been isolated, and German's group determined that Rfx6 was part of the neurogenin 3 network. But Rfx6 stood out: “It is expressed early in the development of the embryo and specifically in the endoderm, the part of the embryo that forms the gut,” says German. “But later in development, Rfx6 expression becomes restricted to the β-cells of the pancreas. And it is never expressed in the brain, as are all the other genes regulated by neurogenin 3.”

When he contacted Polychronakos, German's lab had just knocked out Rfx6 in mice, and the team was anxiously awaiting the results, which “took much longer than they should have”, laughs German. “At one point we started thinking the gene was cursed because everything that could have gone wrong with those mice went wrong.”

But perseverance paid off, and in the end the researchers got their knockout mice, which bore a striking resemblance to the patients Polychronakos had described. In the meantime, Polychronakos's gene-sequencing results had come through, confirming that the faulty gene in his patients was indeed the human counterpart of Rfx6 (see page 775).

Experiments in mice indicate that Rfx6 is one of the earliest genes to be turned on by neurogenin 3 and may thus provide a valuable tool in efforts to generate β-cells for patients with diabetes. “We now have a more complete picture of how to make a normal, mature β-cell,” says German.