Saturday, April 21, 2007

Science retracts major Arabidopsis paper

Four out of five authors of a Science paper that the journal called a "breakthrough of the year" in 2005 have retracted it, saying that the data it was based on could not be replicated.

The study, which described the migration of mRNA to initiate flowering, was based on real-time PCR data, which researchers in the Ume¥ Plant Science Center lab where it had been performed found impossible to replicate. According to principle investigator Ove Nilsson, first author Tao Huang had manipulated data, removing certain points and giving increased weight to others.

Huang, the only author not to agree to the retraction, maintains that the data omissions were valid and documented. "Although I can understand and respect Professor Ove Nilsson... I think the retraction for this paper should not happen, [and was] at least immature," Hwang wrote in an E-mail to The Scientist.

The work has been cited in 54 papers according to the ISI Web of Science. "To some extent, some may have been misled," Takashi Araki, a professor at Kyoto University wrote in an E-mail to The Scientist.

"Obviously these have been very difficult times for me and my group," said Nilsson, "And of course, it's a big set-back for the Arabidposis part of our research." But he says his lab has continued to publish on flowering in trees, and found no errors in other subsequent papers. "People realize that we have handled this according to the books. We had discovered this ourselves and we took the necessary steps to correct it," Nilsson said.

For 70 years scientists have known that a signal, the so called florigen, travels from the leaves, which detect such changes as temperature and day-length, to the growing tip of the plant where it initiates flowering, but they hadn't known what that agent was. In August 2005, Huang et al. reported that mRNA of the flowering locus T, or FT gene, travels to the tips, where it is transcribed into FT protein to signal flowering.

When the paper came out, it was well received. Other studies had already shown that plants used RNA in signaling, so when the paper emerged implicating the FT mRNA as a moving element "it was not that surprising," said Brian Ayre a plant biologist at the University of North Texas. But, he said, this retraction could cause a bit of backpeddling for some in the field, since the FT mRNA theory had already made it into several prominent plant textbooks.

Araki said that he and his group were suspicious of the finding when it was first published in part because his group's results showed that the FT protein, rather than mRNA was actually the moving element and partly because of "very small statistical errors in their real-time PCR data," he said. He also mentioned that "the paper seemed to fit too well with the popular textbook knowledge."

Two papers published online in Science this week, one authored by Laurent Corbesier et al, at the Max Planck Institute for Plant Breeding Research, and another by a Shojiro Tamaki et al, from the Nara Institute of Science and Technology in Japan, indeed indicate that the elusive florigen is the protein product of the FT gene, rather than mRNA. Corbesier's work in Arabidopsis and Tamaki's work in rice both bolster previous work in tomatoes which did not find mRNA, and implicated the potential use of proteins, published in the Proceedings of the National Academy of Sciences by Eliezer Lifschitz.

Nilsson told The Scientist he first discovered an anomaly after Huang, a visiting scientist, left his lab to return to China this past September. Another student and a postdoc, wishing to continue Huang's work found data points marked in red that had been removed from the final analysis. Nilsson's lab was unable to reproduce the data and contacted Ume¥ University which resulted in both an internal and external investigation.

Huang said the removed data points were "irrelevant to the experiment," because variable temperatures of a heating plate rendered certain data points invalid. He said he had circulated his results for review, with the anomalous data points marked in red, before he left the lab. He wrote in an E-mail, "None of my fellow labmates thought it was improper to exclude those irrelevant samples at that time." He also mentioned that he had not been contacted by members of the external investigation.

Lars Rask, a professor at the University of Uppsala who was one of the two people who conducted the external evaluation, confirmed that Huang was not interviewed. Instead, the committee evaluated correspondence between Huang and Nilsson after Huang's departure. In an E-mail to The Scientist, Rask wrote, "As far as we could tell from the E-mails, [Huang] realized that there were potential problems in the experiments carried out."

Despite the controversy surrounding Huang's research, the field as a whole may not be tremendously affected. The original Huang paper had not discounted the possibility of an FT protein playing a role. "That left the door wide open," said Colin Turnbull, a co-author of the Corbesier paper. "All we're trying to do as a community is move forward. And FT is definitely important and the fact that it's a protein is just as exciting as if it had been RNA," he said.