If you need evidence of the value of transparency in science, check out a pair of recent corrections in the structural biology literature.
This past August, researchers led by Qiu-Xing Jiang at the University of Texas Southwestern Medical Center corrected their study, first published in February 2014 in eLife, of prion-like protein aggregates called MAVS filaments, to which they had ascribed the incorrect “helical symmetry.” In March, Richard Blumberg of Harvard Medical School, and colleagues corrected their 2014 Nature study of a protein complex called CEACAM1/TIM-3, whose structure they had attempted to solve using x-ray crystallography.
In both cases, external researchers were able to download and reanalyze the authors’ own data from public data repositories, making it quickly apparent what had gone wrong and how it needed to be fixed — highlighting the very best of a scientific process that is supposed to be self-correcting and collaborative.
According to Eric Sundberg, Associate Professor of Medicine at the Institute of Human Virology of the University of Maryland School of Medicine, the Nature correction (on which he assisted) was an example of “how the process should work.” Specifically:
for both authors and readers to uncover potential problems in papers, and then to work together if necessary or possible to get them fixed.
Sundberg, a structural biologist who specializes in the class of proteins featured in the Nature study, became aware of the Blumberg study when one of his postdocs attended an on-campus journal club that discussed the paper. The “crystallographic statistics,” Sundberg told us, were “worse than expected.” He also knew that one member of the complex, CEACAM1, has a strong tendency of pair up with itself (that is, to “homodimerize”).
Sundberg (and Steve Almo at Albert Einstein School of Medicine, who also questioned the study), “knew, more than we knew, the extremely powerful tendency of CEACAM1 to homodimerize,” said Brandeis University structural biologist Gregory Petsko, who led the Blumberg team’s x-ray work. “And they could tell from the paper that we hadn’t worried about that enough.”
Because authors publishing x-ray structures are required to make their raw data publicly available, Sundberg’s team could download and test them. They quickly realized that what the authors assumed was a blend of CEACAM1 and TIM-3 was, in reality, just two copies of CEACAM1. Sundberg communicated that fact to Blumberg in an email. “Two minutes later he was calling me on the phone,” Sundberg said.
Blumberg told us:
I was floored, to say the least.
He and his team set out to determine whether Sundberg’s critique was accurate, and then to fix the problem. Sundberg and his team helped in the effort, providing new protein expression reagents and protocols to produce better results. “It was a very stressful six-month Paper Chase period where everybody was working 24/7,” said Blumberg, referring to the 1973 law school drama.
In the end, the team never did solve the CEACAM1/TIM-3 structure; instead, they provided new biochemical evidence of the interaction between the two molecules, including both nuclear magnetic resonance and surface plasmon resonance data. The team communicated those new findings to Nature, which effectively re-reviewed it as if it were a new submission, according to Blumberg. Because the remainder of the paper was correct and the lack of structural data did not substantially detract from the team’s overall conclusions, the journal issued a correction; it was published March 16, 2016. It’s too long to print in full here, but here is the first paragraph:
In this Letter, we published the crystal structure of a heterodimer of the human (h)CEACAM1 IgV domain and hTIM-3 IgV domain (Protein Data Bank (PDB) accession 4QYC). Since publication, E. Sundberg and S. Almo have questioned our model, and stated that they had obtained better results refining a hCEACAM1–hCEACAM1 homodimer model against our diffracted amplitudes. We confirm that a homodimer model indeed fits our crystallographic data better, as judged by most statistical measures (see Supplementary Table 1). We have therefore withdrawn the deposited heterodimer model (PDB code 4QYC) from the PDB, and replaced it with a more accurate homodimer model (PDB code 5DZL). We thank E. Sundberg and S. Almo for bringing this to our attention, and apologize for any confusion the original structure may have caused.
“CEACAM1 regulates TIM-3-mediated tolerance and exhaustion” has been cited 28 times, according to Thomson Reuters Web of Science.
Blumberg said his team is now stronger than ever, even counting Sundberg now as a close collaborator. “Eric turned out to be a gem,” he said. As Petsko put it:
I wish we had never had this happen. But if you want something that restores your faith, if you’d ever lost it, in the ability of science to move forward in a good way, this ought to do it — not just for the process, but also for the people. It was a painful but immensely gratifying experience.
The MAVS story was also aided by publicly available data. In February 2014, Edward Egelman at the University of Virginia, an expert in the structural technique known as cryoelectron microscopy (cryoEM), was putting the finishing touches on a manuscript he was coauthoring on the structure and biology of MAVS filaments when he found that he apparently had been scooped. But his apprehension, he says, lasted “about two minutes.”
A quick glance at the published structure suggested the authors had made a mistake, Egelman told us: It was substantially different from his own structure, which had been solved to much higher resolution.
They had imposed the wrong [helical] symmetry.
Egelman voiced his concerns to the editors at eLife, who notified the authors and asked that they make their raw data publicly available and comment on the different structures on their article’s home page. The authors wrote an extensive rebuttal in September 2014, arguing essentially that Egelman’s structure was correct, but not an accurate representation of what the protein looks like in living cells.
When Egelman downloaded the raw data, however, he found that in fact, his structure was a better fit for the authors’ data than theirs was, and he published those observations in an eLife article of his own in December 2014. After Jiang and his team collected new, higher-resolution cryoEM data over the following year, they corrected their original study in August 2015.
According to Wes Sundquist at the University of Utah, who was the reviewing editor on the eLife studies, the peer reviewers who originally vetted the Jiang paper “expressed some concerns that they hadn’t absolutely proven their case that they had the structure correct,” but were ultimately satisfied by the different lines of evidence the authors collected.
Once it was clear the model was incorrect, however, the journal opted to correct the record rather than retract, Sundquist told us, for two reasons:
We felt that the public good was served if they themselves did a reconstruction that proved that their original reconstruction was wrong. And that they had the chance to say the rest of their work was not wrong.
Again, we’d like to print the entire correction notice for “Structural basis for the prion-like MAVS filaments in antiviral innate immunity,” but it’s many pages long. So here’s the first paragraph:
Mitochondrial antiviral signaling (MAVS) protein forms prion-like aggregates mediated by the N-terminal caspase activation and recruitment domain (CARD) and activates antiviral signaling cascades. Purified MAVS CARD from culture cells self-assembles into filaments. Previously, we reported a low-resolution cryoEM structure of MAVS CARD filament, which exhibits a C3 symmetry with a rotation of −53.6° and an axial rise of 16.8 Å for every unit in the filament (Xu et al., 2014). Recently, a cryoEM reconstruction of MAVS CARD filaments at 3.6 Å resolution was reported with a C1 helical symmetry of a rotation of −101.1° and an axial rise of 5.1 Å per subunit (Wu et al., 2014). The differences in these two models were carefully analyzed recently (Egelman, 2014), which suggested that the helical ambiguity in helical reconstruction was not fully resolved in our previous analysis (Xu et al., 2014). We recently collected a new dataset at higher resolutions. Using a newly developed method for analysis of helical filaments (Clemens et al., 2015), we obtained a 4.2 Å resolution reconstruction of MAVS CARD filaments purified from mammalian cells under native conditions. The new model shows that the MAVS CARD filament exhibits a C1 helical symmetry in agreement with Wu et al. (2014).
We contacted Jiang, who told us he didn’t have time to speak in detail about the correction, but noted:
As for transparency, there is not much to discuss because it is expected for every one to do.
“Structural basis for the prion-like MAVS filaments in antiviral innate immunity” has been cited 43 times since it was published in 2014, according to Thomson Reuters Web of Science.
Two years ago, we praised Jiang for “doing the right thing:” In 2013, the researcher retracted a PNAS paper within six weeks of publication after a Nobel Laureate noticed an error.
Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our new daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.