“We were completely shocked:” Plant biologists issue mega-correction

Plant biologists have issued a major correction (what we dub “mega“) after realizing a significant mistake in their experiment.

The 2014 paper shows that a protein known as RAP plays a key role in chloroplast biogenesis. But as Ludwig Maximilians University-based authors Alexandra-Viola Bohne and Laura Kleinknecht continued to do their research, they found an error in the design of primers they used to synthesize the RNA for their experiments — and told us they are concerned other researchers could run into the same problem.

Although the authors considered retracting the paper, since its main conclusion was unaffected, they issued a correction notice, published in April in Plant Cell:

In the course of on-going work, the authors realized mistakes in the design of primers used to generate templates for in vitro transcription of RNA probes by the T7 RNA polymerase. Templates were generated by annealing of primers with incorrectly positioned T7 promoter sequence elements in reverse primers. Therefore, no RNA synthesis should have occurred. However, as observed in native agarose gels as well as in the analysis of synthesized RNAs by RNase T1 digestion, mis-designed primers had a strong self-annealing capacity leading to undefined RNAs of expected sizes. As even correctly designed primers showed self-annealing, new experiments were performed either with PCR products used as templates for in vitro transcription or synthetic RNA oligos.

While the general conclusion on the function of RAP in 16S rRNA maturation is not touched by these errors, their consequence is that the determination of the RAP binding site within the 16S precursor RNA (Fig. 4C) as well as in vitro RAP binding affinities to RNAs (Fig. 6) were not correctly resolved, for which the authors apologize. The corrected experiments do not support binding of RAP to FP1 as stated before. Instead, rRAP showed a higher affinity to the FP2 probe as compared to the two other reported footprint sequences. However, the affinity of rRAP for FP2 seems to be only moderately increased as compared to FP1 and FP3 for which no distinct footprint was detected (corrected Fig. 6, corrected Fig. 4C). Therefore, it is also possible that RAP binds to another sequence within the 16S rRNA precursor or that additional determinants like overall rRNA structure or other trans-acting factors enhance selective binding of RAP to FP2 in vivo. Nonetheless, additional data provided in Fig. 9 support a role of RAP in precise trimming of the mature 16S 5 ’end.

The correction notice also provides six revised figures.

“RAP, the sole octotricopeptide repeat protein in Arabidopsis, is required for chloroplast 16S rRNA maturation” has been cited 13 times, according to Thomson Reuters Web of Science.

In an email signed by both last author Bohne and first author Kleinknecht, they provided us with more details on the error, including a note of caution for others using the same techniques:

In the course of ongoing work, we had decided to further fine-map the binding site of the RAP protein on its RNA target and therefore looked at the original oligo sequences which had been used for the Plant Cell paper.

During this process, we suddenly noticed the fatal mistakes we had made in primer design.  In brief, for the generation of templates for in vitro transcription of RNA probes we had been following a primer-annealing strategy (Williams-Carrier et al., 2008; Ruwe and Schmitz-Linneweber, 2011). Unfortunately, these primers were misdesigned. The major mistake was a switch of 5´ to 3´ directions resulting in false-positioned T7 RNA polymerase promoter elements. We were completely shocked by our findings, since in theory, these misdesigned oligos should not have been able to serve as template for T7 RNAP driven in vitro transcription. However, we constantly generated RNAs of the expected sizes using these wrong oligos. Therefore, we as well as the reviewers of the paper never had a reason to question the design of the oligos and did not notice the mistake beforehand. So, at this point we were left with two burning questions: How can wrong oligos drive efficient in vitro transcription and what is the correct binding site of RAP.

By running annealed oligos on native gels, we quickly figured out that many of them had the ability to self-anneal and, moreover, were sufficient templates for in vitro transcription by the T7 RNA polymerase. Even correctly designed primers showed a strong preference for self-annealing and did result in false RNAs (as tested by RNase1 digestion of generated RNAs). To our knowledge, such artificial T7 RNA polymerase activities have not been reported yet. We feel that experiments involving primer annealing approaches for RNA synthesis should therefore be taken with caution. We have prepared a documentation containing our experimental analysis of error (which did not form part of the recently published Correction in Plant Cell) and we will be pleased to provide these data to anyone upon request.

To elucidate the true binding site of RAP, we used alternative methods that avoid the primer-annealing strategy, we had used before. By now applying RNA gel blot assays, we were able to detect the putative RAP binding site further downstream of the previously published one (Fig. 4C). Moreover, we ordered synthetic RNA oligos to repeat the RNA binding curve experiments (Fig.6).

The authors told us that they considered retracting the publication when they first realized the mistake:

However, with the new information on hand and the fact that the overall RAP story was not affected by the shifted binding site, we decided in accordance with the Plant Cell editorial board (which we had contacted in the meantime) that we submit a Correction. This Correction was subjected to peer reviewing and in the course of the reviewing process additional control experiments were suggested to further substantiate our new findings (e.g. the analysis of the rbf1-1 mutant; Fig. 4B and 4C). While we were working on these controls, we realized that RAP has an additional function in the precise trimming of the mature 16S rRNA. Thus, also these data were included for a comprehensive description of RAP´s function. As a consequence however, the number of new Figures increased and influenced a bit the readability of the entire paper. We apologize for that but we feel a Correction was the best course of action considering that the principle function of RAP in processing of the 16S rRNA was not touched by the discovered mistakes, but the precise binding site had to be reassigned.

We feel that we have tried our best in clarifying the issue and we hope that our mistake has not caused anyone too much trouble.

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.