Retractions unsettle structural bio
Recent findings upend conclusions from five highly-cited papers
Scientists from the Scripps Research Institute led by Geoffrey Chang
have retracted three highly-cited papers (1
) in Science
on ATP binding cassette transporters, stating that new findings by another group invalidated their proposed structure and biology of the transporters. Two more retractions will appear in upcoming issues of the Journal of Molecular Biology
and Proceedings of the National Academy of Science
. The retracted papers span more than five years of work.
"I'm not sure there's ever been a retraction of this magnitude in the field," said Kasper Locher, a researcher and professor at the Institute of Molecular Biology and Biophysics in Switzerland, and an author of a September 2006 paper
that illustrated major problems with Chang et al's findings.
In an interview with The Scientist
, Chang said he regretted any confusion his papers may have caused. "I deeply apologize to those that used the old structures to come up with results that are incorrect," Chang said. "I feel pretty bad about that."
The five retracted papers have been cited 729 times since their publication.
Transporter proteins play a direct role in multidrug resistance, a growing problem seen in antibiotic resistance and in the treatment of diseases such as cancer. The retracted papers focused on the ATP binding cassette (ABC) transporters MsbA and EmrE.
Chang et al's first paper about these transporters, published in Science in 2001, reported the first structure of a bacterial transporter protein using x-ray crystallographic analysis. Chang used this structure and the in-house software that had modeled the data to help gather data for four subsequent studies, published through 2005.
However, some structural researchers reportedly had suspicions
of Chang's structures from the first 2001 paper. Biochemical data suggested a different formation of homologous proteins, but given that Chang et al was the first to image the entire protein there was no visual evidence to which curious scientists could compare the findings, according to Christopher Higgins
, director of the MRC Clinical Sciences Centre at the Imperial College of London, and a researcher of ABC transporters.
Higgins said he had early doubts
about Chang's structures. "We looked at [the structure] and knew it was wrong, but couldn't prove why," Higgins said.
Then, last September, Roger Dawson and Kasper Locher published their paper in Nature
, presenting a new, highly-resolved structure of a homologous transporter, which illustrated an inherent flaw in Chang et al's research. In his retraction notice
, Chang writes that a glitch in his own software had made his structure incorrect, an unintentional error that undid half a decade's worth of work.
The software glitch had switched data, creating an image with the ribboned-strands of the transporter protein in reverse -- essentially a mirrored image of the correct arrangement. Also, the end of the strands, or fingers, were also mis-cast, all by a mere 10 lines of the software program, according to Chang. Unfortunately, at the time of the research, Chang said he considered the software's original interpretation of the data as trustworthy as a social security number. "I didn't question it then," said Chang. "Obviously now I check it all the time," he added.
Higgins said that the erroneous structures have caused a great deal of confusion over the past five years but that the retraction is an affirmation of the strength of the scientific method. "We go into the unknown and gradually bits of data come together," said Higgins. "It's a positive reflection on science."
Links within this article:
G. Chang and C.B. Roth, "Structure of MsbA from E. coli; a homolog of the multidrug resistance ATP binding cassette (ABC) transporters," Science
, September 7, 2001.
.C.L.Reyes and G. Chang, "Structure of the ABC transporter MsbA in complex with ADP.vanadate and lipopolysaccharide.," Science
, May 13, 2005.
O. Pornillos et al, "X-ray structure of the EmrE multidrug transporter in complex with a substrate," Science
, December 23, 2005.
G. Chang, "Structure of MsbA from Vibrio cholera: a multidrug resistance ABC transporter homolog in a closed conformation," Journal of Molcular Biology
, July 4, 2003.
C. Ma and G Chang, "Structure of the multidrug resistance efflux transporter EmrE from Escherichia coli," Proceedings of the National Academy of Sciences, March 2, 2004.
RJ Dawson and K Locher, "Structure of a bacterial multidrug ABC transporter," September 14, 2006.
M. Stephan, "A tale of two transporters," The Scientist
, November 17, 2003.
C. Higgins and KJ Linton, "Structural biology. The xyz of ABC transporters," Science
, September 7, 2001
G. Chang et al. Retraction, Science
, December 22, 2006
January 4, 2007
Few retractions are associated with numerorus avenues of molecular biology, simply because the nebulous hypotheses are never actually tested in the complex systems. In many cases, compensatory systems are the actual cause of a phenotypic change. However, retractions in structural biology are extremely unfortunate! I do not know, or blame, the authors. Problems of this sort are often caused by the necessity of grant applicants to list excessive numbers of publications. I wish that reviewers would actually "read" the articles and not merely count the number of publications. However, I fully understand that the the reviewers themselves are also very busy, and this is often the case for individuals asked to review manuscripts! Errors of this kind would be greatly diminished if quality, rather than quanity, was of greater importance in the evaluation of a publication. Also, we should respect scientific achievement and not "apparent" productivity. This change should be accompanied by the establishment of "scientific impact factors" for individual publications, rather than apply impact factors to specific journals.
January 5, 2007
I agree with the gist of Dr. Sifers comment, but want to emphasize that impact factor does not always provide a clear index of the importance of scientific work. Impact factor can be both falsely low or falsely high. For instance, truely groundbreaking research might go without much response for months to years before others catch up and the work begins to impact the literature. On the other hand, high impact work can (as in the present case) later be shown to be incorrect or even worse, it can become highly sited simply because the work was widely recognized and referenced as being erroneous. Therefore, there is no foolproof alternative to the time-intensive method of reading a scientists papers to judge their relative importance.
January 26, 2007
What a nightmare! For Chang to not benchmark his software against a validated data set is reprehensible. Moreover, for him to continue to publish these structures that were SO OBVIOUSLY WRONG is inexcusable. He should be punished not for the former, but for the latter. Note to Chang-ites: If your structure disagrees with every single shred of biochemical data that comes before it, something is probably wrong about your structure! Many have sufferred by these incorrect structures, and so should he. I for one am happy as can be that he was exposed.