by Andre on 21 May 2008
Given the recent coverage of the retraction of two high profile enzyme design papers, I invited a friend of mine, Ross Anderson, who works in the field to comment on the situation. Here’s what he had to say. -Andre
There’s no fakes like Homme – the headlines and puns write themselves. The award-winning Duke biochemist Professor Homme Hellinga, the darling of protein design and by all accounts a bombastic, ruthless self-promoter has hit some hard times recently.
The current controversy centers around two papers published in Science and the Journal of Molecular Biology in which the program DEZYMER is used to redesign ribose binding protein (RBP), a non-catalytic periplasmic receptor, to incorporate triose phosphate isomerase (TIM) activity common to the TIM-barrel family of proteins. This impressive achievement was augmented by the in vivo recovery of a TIM-knockout strain of E. coli grown on lactate by the redesigned RBP – conditions which require TIM activity. In other words, he and his co-workers converted a non-enzymatic protein into one that not only had an activity he intentionally designed but that could also substitute for the natural enzyme in bacteria. Impressive stuff. Or so we all thought.
After a collaboration was set up between Hellinga and Dr. John Richard at SUNY Buffalo, Richard’s group were unable to reproduce the activity reported in these papers. They discovered that his purification had inadvertently dragged through some wild-type TIM enzyme and that this was responsible for his results. A retraction soon followed from Hellinga and he soon laid the blame on his graduate student who performed the original work, Mary Dwyer. Duke University subsequently found Dwyer innocent of experimental fraud but has yet to formally investigate Hellinga himself. Not a particular surprise, the cynical among us would think, considering the money he has brought to his department.
Worse still, Richard and others claim that even the retraction is dubious and fails to account for TIM activity changes in TIM-defficient E. coli when the designed proteins are mutated to destroy their activity. The implication is that the results are in essence a fantasy, although Dwyer claims that the variability of the assay and protein expression are to blame. Hellinga has so far declined to comment on the recent questions regarding the retraction.
Taken as an honest mistake, this situation is a protein designer’s worst nightmare: publication in a major journal followed by retraction due to protein contaminated with the very wild-type enzyme whose activity was being reproduced. Though as a fellow protein designer, the experimental design seems surprising to me. Why, if a TIM-defficient E. coli strain is already in use for in vivo work, do you use a TIM-expressing strain for your protein expression? Surely contamination is on your mind. Dwyer’s claim of variable protein expression lends a possible explanation but the co-purification of active wild type TIM enzyme is almost inexcusable, especially if your work is to be used as a high profile demonstration of this kind.
So there were mistakes. Dwyer claims that she informed Hellinga about her doubts of experimental validity. That Duke’s investigation exonerated her goes some way to confirm what most of us already know: the ultimate responsibility always lies with the PI. This is not to say that the PI should personally bear the burden of a student who falsified data. What they should do, however, is to frankly own up to the mistakes made and to issue a full and honest retraction.
In the end nobody wins from this situation: Hellinga is currently on a damage control tour and there are rumblings that more of his work should be closely reexamined; Dwyer’s career is forever tainted from the association with the retracted papers and the rumors that inevitably follow; Richard’s lab spent considerable time and money attempting to reproduce bad work and as the Nature editorial states:
Richard’s subsequent efforts to correct the scientific record thus came at considerable cost, with no discernable benefit to his own career.
How much published work is invalid? How many times have scientists attempted to reproduce published data only to find it irreproducible? I sincerely hope that the answers to these questions is a resounding “not very often”. Every scientist has a commitment to honestly present their work and based on this we trust, for the most part, what we read in peer-reviewed journals. Break that trust and you will face the consequences.
A last comment from Hellinga in his own words:
Do you think I’ll be more famous than Darwin one day?
Probably not, Homme.
Dwyer, M. A. , Looger, L. L. & Hellinga, H. W. Science 304, 1967–1971.
Allert, M. , Dwyer, M. A. & Hellinga, H. W. J. Mol. Biol. 366, 945–953.
Dwyer, M. A. , Looger, L. L. & Hellinga, H. W. Science 319, 569. (retraction)