Protein isolation can be tricky in plants, such as thale cress. Credit: J. BURGESS/SPL

Two years ago, three papers1, 2, 3 made a splash in the plant biology world by identifying protein receptors for abscisic acid (ABA), a key hormone in plant physiology. This week, Nature is retracting one of those papers1 after reports that the work could not be replicated. The two other papers still stand, although one3 has been questioned by several researchers4, 5, 6, 7.

The retraction is a setback in the search to find receptors for ABA, a notoriously difficult task. ABA responds to environmental stresses such as drought, and is an alluring target for agricultural companies hoping to produce drought-resistant crops.

The retraction in Nature comes from the team that did the work, led by Robert Hill of the University of Manitoba in Winnipeg, Canada. Hill says the experiments were carried out by postdoc Fawzi Razem, who has since left the university; Hill says he does not know for where. A Nature reporter was unable to contact Razem.

John Danakas, director of public affairs for the university, says that he cannot comment on this case in particular, but says that "normally when there are problems with research results being reproducible, there would be an investigation at the senior administrative level".

According to the search tool Scopus, Hill's paper has been cited 120 times, and is the most highly cited study among the 95 results for 'abscisic acid receptor' in the past three years.

Hill says he moved on from the project after the paper was published, but started to realize something might be wrong after some of his postdocs returned to it in late 2007 and could not replicate the results. He was also contacted by Catherine Day of the University of Otago in Dunedin, New Zealand, who could not repeat the work either. "I suspect it has hurt a lot of people," he says. "It certainly hurt my lab quite significantly. I certainly feel very bad about it and am very apologetic to the community.

According to Hill, his team interpreted an experiment as showing that ABA broke up a binding interaction between the receptor candidate FCA, involved in flowering, and another protein, FY, because ABA bound better to FCA than did FY. In fact, he says, the FY protein was not properly prepared, and probably never bound to FCA at all. Another experiment purporting to show direct binding between FCA and ABA was miscalculated and, says Hill, "the assay procedure is also suspect".

Some plant-hormone specialists had wondered about the results of this and the other papers from the beginning, as they did not fit with the rest of what was known about ABA. The studies all used in vitro biochemical techniques to identify the receptor, rather than the more usual method of identifying and cloning mutated genes before determining if they coded for a receptor.

"There was nothing really wrong with the paper; you couldn't say 'this gel isn't any good'," says Peter McCourt, an ABA specialist at the University of Toronto, Canada. But "none of it linked up with any of the genetics that had been done in the previous 20 years".

The other papers identify similarly unexpected receptors. "People who are doing this are sceptical and are not putting all their money on these findings," says Jianhua Zhang, who works on other aspects of ABA at Hong Kong Baptist University.

Day blames the problems on tricky assays. "I think they've just had a bad start. These proteins are difficult to work with." McCourt thinks such experiments might be too complicated to pursue. Collecting proteins from plant cells for binding studies is easier said than done, he says, as plant cells are particularly crowded, complex places.

Meanwhile, Hill says, "I just want to get this behind me."