“[T]hese things can happen in every lab:” Mutant plant paper uprooted after authors correct their own findings

Three biologists at Tokyo Gakugei University in Japan have retracted a 2014 Frontiers in Plant Science paper on abnormal root growth in Arabidopsis “in light of new experimental evidence” showing they fingered the wrong mutant gene. The journal editors are hailing the retraction as an “excellent example of self-correction of the scientific record.”

The paper, “Mechanosensitive channel candidate MCA2 is involved in touch-induced root responses in Arabidopsis,” described the abnormally behaving roots of a mca2-null mutant Arabidopsis plant.

A subsequent string of experiments by the same research team—including DNA microarrays, RT-PCR, and a PCR-based genomic deletion analysis—demonstrated that two other mutations that somehow creeped into their experimental populations may have been to blame for the abnormal root behavior.

It’s a notably thorough and informative retraction notice from Frontiers, an open-access publisher with a history of badly handled and controversial retractions and publishing decisions. The notice describes the new experiments and the previous, erroneous results:

The authors and the journal wish to retract the 21 Aug 2014 article cited above in light of new experimental evidence.

Following the publication of the study, we performed a DNA microarray analysis to detect genes with expression levels that were specifically lower in the mca2-null mutant than in the Col-0 wild type, and found that the expression level of the AXR4 gene (At1G54990), which encodes a protein required for the subcellular localization of the auxin influx carrier AUX1 (Dharmasiri et al., 2006), was significantly lower in the mca2-null mutant. To confirm this finding, we then performed a semi-quantitative reverse transcription-PCR analysis using the primers axr4-f1 and axr4-Cr1 (Figure 1A), and found that the RT-PCR product was detectable in some mca2-null seedlings at wild-type levels but not in other mca2-null seedlings at all. This result suggested that some mca2-null seedlings have a certain lesion in the AXR4 locus, and a PCR-based genomic deletion analysis (Figures 1B,C) followed by DNA sequencing confirmed this speculation. Our conclusion is that most of the mca2-null seedlings used in the study presented in the above paper had a homozygous 2592-bp deletion that started from the intron between exons 1 and 2 of the AXR4 gene and reached the intron between exons 1 and 2 of the adjacent gene AT1G55000 (Figure 1A). Therefore, the phenotypes presented in the above paper may be ascribable to the axr4 mutation, the at1g55000 mutation, or both or even triple mutations, but not to the mca2-null mutation. The AT1G55000 gene encodes the peptidoglycan-binding LysM domain-containing protein involved in a macromolecule catabolic process in the cell wall.

Our phenotypic study revealed that none of the mca2-null AXR4⁺ AT1G55000⁺ seedlings showed all the abnormal phenotypes reported in the above paper, regarding the skewing, waving, and bending responses of the root. In contrast, the seedlings of the axr4-1 (Ws-2 background) and axr4-2 (Col background) single mutants obtained from the Arabidopsis Biological Resource Center (ABRC germplasm names CS8018 and CS8019, respectively) showed the same abnormal phenotypes as those described in the above paper. We also confirmed that the abnormal phenotypes for the skewing, waving, and bending responses of the mca2-null arx4 at1g55000 triple mutant were identical to those of theaxr4-1 and axr4-2 single mutants. These findings clearly demonstrated that the abnormal phenotypes described in the above paper were ascribed solely to the mutation in the AXR4 gene.

The notice goes onto speculate how the arx4 at1g55000 mutants got into the study. According to senior author Hidetoshi Iida, the team had never used axr4 at1g55000 mutant plants in the laboratory before the publication of the paper. It may be that the unexpected mutations had been present in heterozygous pairs in some seed stocks they used, Iida hypothesizes, and during self-pollination (a common technique for maintaining Arabidopsis in the lab),

it is possible that the homozygous axr4 at1g55000 mutation spread in some of seed stocks.

Iida suggests ways for other scientists to avoid similar errors when growing and pollinating plants in the laboratory:

I would recommend every researcher to use at least two independent lines or ecotypes with a mutation in the same gene of interest. The time may come in the near future when every researcher will have to check the full nucleotide sequence of her/his own Arabidopsis mutant with the advancement of DNA-sequencing and data-analyzing technologies.

Prior to the retraction, Iida says he sent a note about the error to some 50 researchers around the world to whom he had sent the seeds of the mca2-null mutant. Their reactions suggest similar errors in plant genetic studies may be common. Iida wrote:

Soon after that, to my surprise, many of the researchers sent me warm and considerate messages. I was very impressed and encouraged by the response of them. In addition, more surprisingly, some of researchers gave me an unexpected comment. The followings are examples:

“These things happen quite often in the phenotypic analyses of mutants.”

“I think these things can happen in every lab.”

“I understand that this kind of problems happens more and more frequently as sequencing and genetics is getting close to saturation.”

“There are many others who would not do the nice and right thing that you did.”

“I suspect there are many labs that have ignored this and moved on.”

These comments suggest that this kind of problems has happened for years but has not been necessarily disclosed to the scientific community. I think, however, that it is important for the community not to blame the researchers who hesitate the disclosure of the problems but to make an atmosphere that encourages them to disclose the problems. I believe that this atmosphere should not only be encouraging but also be accepting and acquitting in case of unforeseen problems because this kind of problems must be shared by the community.

Editors at the journal, Steven Huber, Andreas Weber, and Frederick Fenter, lauded how the authors handled the error in a statement they sent us:

In February of this year the handling editor of the manuscript, Dr Gabriele Monshausen, approached the Frontiers editorial office and the chief editors of Frontiers in Plant Science / section Plant Physiology following communication with the authors—in particular the senior author of the study, Dr Hidetoshi Iida. The authors had in fact discovered that the genotype of their Arabidopsis strains was not as expected, and had carried out confirmatory experiments that led them to doubt their previous data. When they contacted Dr Monshausen, the authors did so with proof that their strains contained a second genetic lesion that called into question the conclusions they had drawn in their Frontiers article. 

The chief editors, Dr Steven Huber and Dr Andreas Weber (both signatories of this letter), agreed with Dr Monshausen that the authors had acted very responsibly in disclosing this issue; and that they had even performed a real service to the community, as their history with the strains, which they describe in their retraction notice, suggested that the seed stocks originally obtained from the Arabidopsis Biological Resource Center were not of the expected genotype. The chief editors hope that the retraction notice will serve to alert the community to possible artifactual results obtained from other strains of similar provenance, and feel that this retraction is an excellent example of self-correction of the scientific record.

Like Retraction Watch? Consider supporting our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, and sign up on our homepage for an email every time there’s a new post.