BioEd Online

A retraction of a paper by a high-profile chemist has left a prize on the table in the protein-design field, and now, another scientist is poised to grab the limelight.

On 1 February, Homme Hellinga and two former students retracted¹ a June 2004 paper that had been lauded as a breakthrough in the field of protein design². A mistake in the experimental procedure meant that their conclusion wasn't valid. But the work served as a catalyst for others to follow the same path, and now another scientist, who says that he was partly inspired by Hellinga, has accomplished the feat Hellinga thought he had achieved.

In the now-retracted 2004 paper, Hellinga’s team at Duke University in Durham, North Carolina, claimed to have used computer programming and directed evolution — a method for making and screening proteins with desired properties — to transform a humdrum protein into a powerful enzyme for catalyzing biological reactions.

The work was heralded because scientists had never before designed and built a highly active enzyme out of a protein with no enzymatic activity. The feat is prized for its potential practical applications, and because it is one of the ultimate tests of our understanding of how enzymes work.

The paper excited the protein design field, inspiring other scientists, including David Baker, a biochemist at the University of Washington in Seattle, to explore its possibilities. Baker took what he saw as the next step: trying to design an enzyme with a novel function, which works differently from any naturally occurring enzymes. Now, Baker has a paper in press describing his successful method — and so he is likely to claim the plaudits that Hellinga and his colleagues originally took.

Innocent mistake

The mistake in Hellinga's 2004 paper was picked up by John Richard, a chemist at the State University of New York in Buffalo. Richard studies the enzyme triose phosphate isomerase, whose activity Hellinga thought he had engineered.

In 2006, Richard asked for the materials needed to make Hellinga’s engineered triose phosphate isomerase. Richard manufactured the engineered enzyme in Escherichia coli bacteria and then performed a routine purification step to pull the enzymes out of the bacteria. But when he did this, the manufactured enzymes didn’t work — they had no triose phosphate isomerase activity.

With further experiments, Richard found out why: he and Hellinga had used slightly different purification methods. While both methods are standard, Hellinga’s method did not separate E. coli’s natural triose phosphate isomerase from the engineered version. So Hellinga’s analyses were contaminated by the natural enzyme, and that fooled the group into thinking that the engineered enzyme worked. “It was clear that the expressed protein had no activity at all, and that all the activity was from the wild-type E. coli protein,” Richard says.

Last year, Richard notified Hellinga. Soon after, in September 2007, Duke began an inquiry into allegations of scientific misconduct against the student who had performed the bulk of the work, Mary Dwyer. But the inquiry never proceeded to a full investigation; on 4 February, Dwyer was notified that she had been cleared. She has moved on to a postdoctoral fellowship in another laboratory at Duke and says that the episode has been very difficult: “This whole thing has been horrible for me,” she says. “It didn’t have to be handled that way.”

Hellinga accepts that Dwyer did nothing wrong — his lab had long relied on this purification method. “A mistake was made, and nobody caught it — including myself,” Hellinga said. “We were concerned it might not have been innocent, but it was.”

On to something

The silver lining is that Baker's work, soon to be published, may prove it is possible to design working enzymes, as Hellinga set out to do.

Another prominent biochemist, Steve Mayo of the California Institute of Technology in Pasadena, says that his lab has already vetted Baker’s designed proteins: “We’ve tested [Baker’s] materials ourselves and we’ve confirmed that they work, so our confidence level in the work is very high,” Mayo says.

Mayo adds that the episode may hold lessons for the field as a whole: “It will be very useful for the community to have some internal verification before submitting any high-profile result.”

But for Hellinga, this silver lining is bitter-sweet. “The good news is the general ideas are correct; the bad news is that we made a mistake in proving that experimentally,” Hellinga says. “It’s very painful.”