Last week, two papers in Science reported the discovery of florigen, a long-sought compound with the power to make flowers bloom. But if the celebration of its discovery seems a little muted, it is because many researchers have heard this claim before. And this time, the reports come as an old one is retracted amid charges of data manipulation.

The discovery of florigen was heralded in 2005, when another Science paper1 claimed that it was the RNA produced by a gene called FLOWERING LOCUS T, or FT. But now the authors of that paper have retracted their findings, and in its stead come two papers that say florigen is not FT RNA, but the protein produced by the FT gene.

The finding could solve a riddle that has been around since 1865, when German botanist Julius von Sachs observed that illuminating a single leaf on a darkened morning-glory plant was enough to prompt the plant to bloom. That suggested that a signal travelled from the leaf to the site of flower initiation. Some 70 years later, the signal was christened florigen by the Russian plant physiologist Mikhail Chailakhyan.

The hunt was on, and physiologists spent decades testing compounds extracted from flowering plants, only to fail to find the compound responsible. Over time, the florigen concept fell out of fashion, giving way to a hypothesis that the flowering signal was not a dedicated compound but rather a complex mix of nutrient and hormonal signals. “For a long time, florigen was the f-word,” says Joe Colasanti, a plant biologist at the University of Guelph in Ontario, Canada. “You didn't want to bring it up.”

But the recent work shows that researchers were looking in the wrong places over all those years, says Jan Zeevaart, an emeritus plant biologist at Michigan State University in East Lansing. Most people expected florigen to be a small chemical compound. “They weren't looking for proteins,” he says.

Then, in August 2005, two papers reported that although the FT gene produces RNA in the leaf, the encoded protein acts in the tip of the shoot, where flowers form. The simplest explanation was that a product of FT — either the RNA or the protein — somehow travelled from the leaf to the shoot tip. Within a month, a team led by Ove Nilsson of the Swedish University of Agricultural Sciences in Umeå announced that this product was FT RNA. Although that didn't rule out the possibility that the FT protein also travelled from leaf to shoot, it suggested that FT RNA was at least a component of florigen. The discovery was significant enough to make Science's list of 'Breakthroughs of the Year'.

Root cause: the signal for flowering in the thale cress Arabidopsis has been found to be a protein. Credit: J. BURGESS/SPL

But now Nilsson has retracted the paper and has accused Tao Huang, the paper's first author, of manipulating data. Nilsson says Huang selectively excluded some data points and statistically overweighted others. Huang, who left Nilsson's lab for a faculty position at Xiamen University in China after the paper was published, maintains that excluding the data was justifiable. He says he circulated the data — with excluded data points marked — to his lab colleagues before publication, but no one objected to the exclusion. Huang has not agreed to the retraction, calling it premature.

Nilsson's retraction was published at the same time that Science released the two new papers2,3. Both papers — one on the thale cress Arabidopsis and the other on rice — report that although FT gene expression is restricted to the leaves, the protein can travel to the tip of the shoot. And both papers fail to find evidence for movement of FT RNA.

The timing of the papers, coupled with what several researchers have described as an unusually short, 40-day review, has led some to speculate that the papers were pushed to publication more quickly to coincide with the retraction. But Katrina Kelner, Science's deputy editor for life sciences, says the review period was not abbreviated. “It was sensible to have them come out at the same time for maximum clarity of the literature,” she says. “We coordinated them, but the review process of those two papers was in no way abnormal.”

The new work comes with its own share of caveats. Both groups rely on commonly used but indirect measures of protein movement, and some researchers have pointed out that key controls are lacking.

Overall, however, many experts say the new papers are convincing. “None of these is really the killer experiment,” says Detlef Weigel of the Max Planck Institute for Developmental Biology in Tübingen, Germany. “But I would say the overwhelming evidence is that the protein moves.” Zeevaart goes even further. “The problem is solved,” he says.

But with a history as chequered as florigen's, not everyone is ready to close the book. “It's always good to be cautious,” says Colasanti, “especially in this field.”