Hellinga has been under investigation for possible research misconduct, following the retraction of a Science paper on computational design of enzymes in February 2008. This week in the Proceedings of the National Academy of Sciences, Hellinga's former postdoc Birte Höcker and colleagues at the Max Planck Institute for Developmental Biology in Germany dispute the conclusions of his studies on ligand-binding proteins, which appeared in Nature in 2003 and PNAS in 2004. The interactions between ligands and their receptor are central to many biological processes and can potentially be tweaked to design novel biosensors and enzymes.
The original papers from Hellinga's group purported to show that his team modified several proteins from E. coli in order to accept, for instance, the explosive TNT, instead of the sugars to which they normally cinch themselves around. "We applied our algorithms, took the plunge, and built the predicted mutations," Hellinga told The Scientist in an interview last year. "That was very exciting."
Hellinga's group had relied on their in-house fluorescent probe to indirectly report successful binding, but Höcker's group has now used direct measurements based on nuclear magnetic resonance imaging and other techniques to probe five of Hellinga's designed proteins. They found that none of the designed proteins were bound to ligands as expected. The proteins are supposed to change their three-dimensional shape when the ligand binds to their receptor, but the one protein for which they solved the structure remained in its open conformation. Höcker also told The Scientist that her unpublished experiments with commercially available fluorophores did not show any binding, but she did not replicate Hellinga's studies with his custom fluorophores. In general, she notes, the fluorophore assay may be prone to false positives.
Hellinga did not respond to specific questions about the newest paper but provided a statement to the The Scientist, describing several possible reasons for the discrepancy that his group would investigate. "If these studies also show that our original interpretations are in error, we deeply regret that our reports of these designed receptors do not live up to closer scrutiny. In this case, we offer our sincere apologies to researchers whose work was negatively impacted by these reports," according to the statement. (The 2003 Nature paper has been cited nearly 300 times.)
Loren Looger, a coauthor on the disputed papers and now a researcher at the Howard Hughes Medical Institute Janelia Farm campus, made it clear that Hellinga was not speaking for the group and considered the fluorophore assay a dead-end. "I do not dispute anything in Dr. Höcker's work; in fact, I strongly encouraged her to pursue this line of research," he said. He explained that he had encountered problems with the stability of the proteins and the robustness of the assay during his time at Duke and with additional research in his postdoctoral laboratory. He did point out, however, that TNT-binding proteins cloned into bacteria were still able to perform their function in a signal transduction cascade that depended on ligand binding. "To me, this remains the biggest unanswered question of the affair," he said.
This is not the first time that a structural study has contradicted Hellinga's ligand-binding assay. In 2005, Patrick Telmer and Brian Shilton of the University of Western Ontario in London published a paper in the Journal of Molecular Biology showing that Hellinga's zinc biosensor -- a modified maltose-binding protein described in a 2001 PNAS paper -- was not changing conformation as Hellinga's group had claimed, suggesting the protein was not functioning properly. "All of our results mesh with the behavior [Hellinga's group] describes," Shilton told The Scientist, "but then we actually looked at the structure it wasn't closing up the way it was designed to."
And while no one has alleged misconduct with respect to these three papers, questions linger about the earlier retractions. Hellinga was the senior author of a celebrated Science paper in 2004, claiming his team had been the first to design and synthesize a novel enzyme, called novoTIM. It was a scientific first, but after an independent researcher was unable to replicate his results, Hellinga retracted it along with a follow-up paper in the Journal of Molecular Biology. Initially, he directed the blame at his student, lead author Mary Dwyer, who Duke investigated and cleared of research misconduct. Then, as his own behavior came under scrutiny from his critics, he publicly invited Duke to begin an investigation of his role in the disputed work.
Duke representatives said that the investigation is still ongoing. "It would be inappropriate for Duke to comment on any specific proceedings due to confidentiality and other restrictions," Doug Stokke said. Hellinga told The Scientist that the Institute for Biological Structure and Design that Duke had established for him and his wife, Lorena Beese, in 2005 no longer exists.
David Baker of the University of Washington at Seattle has since laid claim to the first designed enzyme, but it has a much lower binding affinity than natural enzymes, which is what inspired Höcker to take a closer look at the problem. "Ligand-binding design has not been attempted [since] the [Hellinga] publications in Nature and PNAS," Höcker said. "We need to focus again on binding in order to improve receptor design as well as enzyme design."
Correction (October 14): The original version of the article incorrectly stated that the zinc biosensor did not bind to its ligand. In fact binding was observed, but it did not adopt the predicted shape change, suggesting it was not functioning properly. The Scientist regrets the error.
Related stories:
[25th July 2008]
[26th March 2008]
[July 2006]
[25th June 2004]
[Comment posted 2009-10-22 13:28:56]
In one section of the remaining questions, Looger does make this statement, which may shed a bit more light on the Nature News comment:
"There were a number of instances in which a given designed protein did not seem to bind its target ligand; these were not reported in the paper. Nonpublication of negative results is extraordinarily common in science; this did not seem inappropriate. The apparent failure rate was not such that it clearly pointed to any
problems with the work."
Alison McCook
Deputy Editor
[Comment posted 2009-10-22 12:48:30]
I would like to read the context for his infamous Nature News article quote:
"'[T]here were a number of instances in which a given designed protein did not seem to bind its target ligand; these were not reported in the paper,' [Looger]adds. 'Non-publication of negative results is extraordinarily common in science; this did not seem inappropriate.'"
[Comment posted 2009-10-20 14:20:08]
Alison McCook
Deputy Editor
[Comment posted 2009-10-20 11:56:27]
[Comment posted 2009-10-16 15:27:27]
Need we ask anyone to tell us these things?"
-Robert M. Pirsig (?Zen and the Art of Motorcycle Maintenance?)
[Comment posted 2009-10-15 03:12:20]
Homme Hellinga and many others who wish to design proteins with specific, desired functions - from the ground up, i.e., from scratch - are just too ambitious, and in too much of a hurry; they seem to place the glory that comes from their work ahead of the truth which must be the cornerstone of all reliable science.
What I find amusing about all this is the ease with which the world of science - and the big journals - managed to allow them to publish such big claims without even requiring them to publish a crystal structure of an engineered re-designed protein. This just goes to show that many reviewers and editors put the 'form' of a paper ahead of its 'content' (and the 'form' here includes the lead-author's towering reputation). This is the failing of much of modern science. I ask : How come the reviewers and the editors failed to realize that man simply doesn't yet understand enough about the sequence-structure relationship in proteins to perform such ambitious 'protein design', i.e., if indeed any of these reviewers or editors understand the complexities of protein structural engineering ?
My group holds that all that man can do - even now - is to learn from nature, i.e., to attempt 'protein re-design' rather than de novo 'protein design'.
At the end of this message are given three recent examples (published papers) of what may be considered by some to be ambitious experiments in protein re-design that have come out of our laboratory (www.guptasarmalab.in). These papers explore, in our opinion, the limits of what is now possible with proteins. Perhaps the future holds more portents.
In all instances we have found that one gets away with making profound changes only when one uses what nature has already developed, in a homologous protein with a superimposable backbone. Otherwise, there is just no way of hoping that the chain will fold as one wants it to.
Oh (and by the way), all of our work is supported by proof of purification of the engineered protein, proof of extensive structural-biochemical characterization by various forms of spectroscopy, spectrometry and biomolecular separation techniques and (in some instances also) by structure determination (i.e., a crystal structure to prove that our 're-design' worked).
We don't have the towering reputations of the Hellingas of this world. So, of course, we get to publish only in BBA and FEBS J. et cetera rather than in Science or Nature.
But we laugh at what is happening to Science, JMB and some other journals, when they publish claims about engineered proteins that have not even been actually purified by the authors!
How much more funny can science-publishing get?
Anyway, those who are interested can check out the following papers in protein re-design, before they decide to debunk all protein design:
[1] Kapoor et al. (2008). Replacement of the active surface of a thermophile protein by that of a homologous mesophile protein through structure-guided ?protein surface grafting?. Biochim. Biophys. Acta : Prot. Proteom. 1784, 1771-1776.
[2] Kapoor et al. (2009). Creation of a new eye lens crystallin (Gambeta) through structure-guided mutagenic grafting of the surface of βB2-crystallin onto the hydrophobic core of γB-crystallin. FEBS Journal 276, 3341-3353.
[3] Kapoor et al. (2009). A functional comparison of the TET aminopeptidases of P. furiosus and B. subtilis with a protein engineered variant recombining the former's structure with the latter's active site. Enzyme Microb. Technol. (in press).
[Comment posted 2009-10-13 20:17:38]
The activity!
Without functional assays, discussions based on detected ligant binding could create more problems.