Two detailed retraction notices appear in PNAS

We’ve fallen a bit behind in our coverage of retractions in the Proceedings of the National Academy of Sciences (PNAS), so we wanted to call attention to two very helpful ones from recent months.

Here’s one notice, which appeared online on August 5:

Retraction for “Identification of an RNA-dependent RNA polymerase in Drosophila involved in RNAi and transposon suppression,” by Concetta Lipardi and Bruce M. Paterson, which appeared in issue 37, September 15, 2009, of Proc Natl Acad Sci USA (106:15645–15650; first published September 1, 2009; 10.1073/pnas.0904984106).

The authors wish to note the following: “In our article, we reported that recombinant subunit 1 of the Drosophila elongator complex, D-elp1, had RNA-dependent RNA polymerase (RdRP) activity and was involved in RNAi, transposon suppression, and endo-siRNA production, but not miRNA targeting. RdRP activity was identified using three assays to interrogate the reaction products: Dcr2 digestion, RNase sensitivity, and nearest neighbor analysis. Although we do see differential Dcr2 cleavage and RNase resistance, the gold standard for second strand RNA synthesis is nearest neighbor analysis, as described previously for the Neurospora crassa RdRP, QDE-1 (1). Subsequent studies revealed that the nearest neighbor analysis was misinterpreted because of two factors unknown to us at the time: First, T7 run-off transcripts used as single-stranded RNA templates have heterogeneous 3′ termini (2, 3); Second, the Flag-tag affinity purified recombinant D-elp1 protein preparations used in our study contain a ribonucleotide terminal transferase activity able to catalyze the addition of single a-labeled ribonucleotide triphosphates to the 3′ hydroxyl terminus of the template RNA. Therefore, nearest neighbor analysis measured the heterogeneous 3′ termini of the template RNA and not second strand RNA synthesis, as initially reported. Given this result, we wish to rescind the interpretation that D-elp1 can be considered as an RNA-dependent RNA polymerase. The basis for the ribonucleotide terminal transferase activity is under investigation. The interpretation of the results does not detract from the fact that D-elp1 is involved in RNAi, endo-siRNA production, and transposon suppression, but not miRNA targeting as shown in Figs. 3 and 4. By clarifying this issue, we hope to promote further productive investigation into the role of elongator subunit 1 in RNAi and transposon suppression. We apologize for any difficulties our original interpretation may have caused other investigators and hereby retract the paper.”

The paper has been cited 31 times, according to Thomson Scientific’s Web of Knowledge. In a story about the retraction, Biotechniques noted that the study had been dogged by questions:

“This paper was received with some suspicion, but there are some important findings like its role in RNAi,” said Luc Swevers, a scientist at the National Center for Scientific Research Demokritos in Athens, Greece, who studies small RNA pathways in insects. “The existence of a RdRP in insects would indicate that there could be amplification of the RNAi response.”

Given all of that, Lipardi — who worked for Paterson at the NIH until joining Merck recently — should get credit for two things:

• One, this is a very detailed retraction notice. There’s really no doubt that this was an honest error.
• Two, they seem to have done everything they could to try to prove their results wrong the first time around — a hallmark of good science — and yet still had to retract the paper. That can’t have been easy to swallow.

Here’s the other notice, which appeared online on August 4:

Retraction for “Phospholipid bilayers are viscoelastic,” by Christopher W. Harland, Miranda J. Bradley, and Raghuveer Parthasarathy, which appeared in issue 45, November 9, 2010, of Proc Natl Acad Sci USA (107:19146–19150; first published October 25, 2010; 10.1073/pnas.1010700107).

The authors wish to note the following: “We have determined that the conclusions stated in our paper are an artifact of errors of analysis. Our methods involved determining the rheological properties of lipid bilayers from analysis of the trajectories of lipid-anchored particles. Errors related to the particle-tracking methods we used generated an artifactual elastic-like signature.

“The imprecision in tracking resulted from determining the position of each particle by fitting the logarithm of the particle’s intensity profile to a quadratic function. Directly fitting the intensity to a Gaussian form is considerably less sensitive to noise and signal intensity. This improved fitting reduces positional uncertainty in test images from around 30 nm to a few nm and, applied to our membrane data, completely eliminates the ‘elastic’ response. Moreover, adding 30 nm of random error to precisely tracked images reproduces the reported elastic signature. Therefore, our data indicate that lipid membranes show a purely viscous character over the entire frequency range examined. Our initial assessments of particle tracking algorithms were not sufficiently stringent, leading to an overestimate of our tracking precision and the use of inadequate fitting methods.

“The artifactual elastic response we reported showed a feature at the chain ordering transition temperature of the lipid membranes examined. We are not certain what aspect of the images, when imprecisely tracked, led to this feature. We suspect that motions of the particles perpendicular to the membrane plane may be responsible.

“The conclusions of the paper are wrong and, regrettably, we must retract the work. We sincerely apologize for any confusion that our report may have caused.”

The paper has been cited five times.

PNAS has retracted five papers this year, most recently one by Anil Potti and colleagues.

Hat tip: Mico Tatalovic