‘This has been a nightmare’: One paper was retracted. The other still lingers.

On a Saturday last November, Philip Tsichlis of The Ohio State University received an email no researcher wants to get. 

Another scientist had tried to replicate a finding in a recent paper of his, and couldn’t. “We believe that our results should lead to some revision of the model you propose,” stated the email, which was released to us by OSU following a public records request. 

It turned out that was an understatement. The email eventually led Tsichlis to discover data fabrication in that paper and a related article. Within a week, he requested the retraction of both papers, one in Communications Biology and the other in Nature Communications, both Springer Nature journals. One was retracted in December, but not the other.

In an email to a Nature Communications editor on November 22nd, Tsichlis wrote: 

This has been a nightmare and I blame myself for not having detected it earlier. However, we cannot go back. I hope that we will retract this paper as soon as possible. 

Seven months later, it remains unflagged. 

The email that spurred Tsichlis to reevaluate his lab’s papers came from Alexandre Maucuer of INSERM on November 13th. 

It referred to an article his group had published in Nature Communications in July, titled “AKT3-mediated IWS1 phosphorylation promotes the proliferation of EGFR-mutant lung adenocarcinomas through cell cycle- regulated U2AF2 RNA splicing,” and was addressed to Tsichlis and the paper’s first author, Georgios I. Laliotis. 

Maucuer, who didn’t respond to our request for comment, wrote: 

We were very interested by your results but still surprised that U2AF2 would experience alternative splicing leading to the expression of an isoform lacking its RS domain. We tried to detect this isoform by analysing a variety of RNAseq data but could not detect the skipping of U2AF2 exon 2. We believe that our results should lead to some revision of the model you propose concerning the action of IWS1 on splicing and cell proliferation. In order to provide this new information to the research community we propose to publish our results as a “matters arising” manuscript in Nature communications [sic] and we are asked to first ask you for an informal reply. You will find our manuscript attached.

A few hours later, Tsichlis responded: 

Thank you for your Email. I understand that your bioinformatics analysis of both our RNASeq data and other publicly available datasets did not detect the alternative RNA splicing of the U2AF2 exon2. To answer your comment, I will have to discuss this issue with our bioinformaticians. However, in addition to the bioinformatics analysis, Dr Laliotis detected the alternative splicing of exon 2 in numerous wet lab experiments, which cannot be ignored. At this point, I need to do two things. First, as I said, I will have to discuss with our bioinformaticians the nature of their analysis and how it relates to your analysis. In addition, I will have to repeat the experiment on the U2AF2 exon 2 splicing myself and I will have another person in the lab to repeat it as well. I believe that the repeat of the experiment is the most important and the most definitive way to answer your comment. Since I will have to repeat everything from scratch, this will take approximately two weeks. Once I repeat it, I will send you all the experimental details, so that you can repeat it as well. I hope that you can wait until all this is done before you submit your paper.

But before he had begun repeating the experiments, Tsichlis realized there was a bigger issue. 

On the Friday after he received Maucuer’s email, Tsichlis emailed the editor of a related paper his group had published in October in Communications Biology, requesting to retract the article and explaining what had happened: 

It is very difficult for me to write this letter. However, I am obliged to report that we have confirmed several irregularities in the data on the alternative RNA splicing of U2AF2, which is a central point in the proposed mechanism of RNA transport we reported in in our recent paper in “Communications Biology” (Phosphor-IWS1-dependent U2AF2 splicing regulates trafficking of CAR-E-positive intronless gene mRNAs and sensitivity to viral infection).

The reason we went back to investigate this issue was an Email from Dr Alexandre Maucuer in INSERM, who indicated that he could not find evidence for the alternative splicing of the U2AF2 exon 2 in several bioinformatics analyses he performed. In the process of addressing his comment, we (me and other people in the lab) realized that the alternative RNA splicing of U2AF2 could not give a functional U2AF2 protein and therefore it could not be a physiological mechanism of U2AF2 regulation. In addition, we carefully scrutinized figure 1g of the earlier paper we had published in “Nature Communications” and we discovered evidence of data manipulation (see attachment). I was planning to repeat the U2AF2 alternative splicing experiment myself, but given this evidence, I decided that it will not be necessary to do that. The problem is that the alternative splicing of the U2AF2 exon 2 is repeated over and over again, and it is a central piece of the model we proposed in the “Communications Biology” paper. How these experiments were done to reproduce an event which cannot happen, is unknown. Based on this, we need to retract the paper effective immediately.

He attached slides that detailed his findings of data manipulation: 

Tsichlis noted that he had informed the OSU office of research compliance about the retraction, and asked to withdraw a third paper with the same first author that was under review at Communications Biology: 

I sincerely apologize for the time and effort you spent reviewing this paper. I am really embarrassed by this, which is definitely a new experience for me. 

The following Tuesday, an editor at Communications Biology responded to Tsichlis with proposed text for the retraction notice, and within hours he had forwarded it to the paper’s other coauthors for their agreement. 

The published notice was identical to the text the editor had proposed: 

The authors are retracting this Article as irregularities were found in the data that indicate the splicing of the U2AF2 exon 2 does not occur as reported in the Article. The irregularities call into question the conclusions and undermine our full confidence in the integrity of the study. The authors therefore wish to retract the Article.

All of the authors agree with the retraction.

The retraction ran on December 15, less than two months after the paper had been first published, and less than one month since Tsichlis contacted the journal to request the retraction. 

The records we received from OSU do not include the original email from Tsichlis requesting to retract the Nature Communications paper, but they do include an email he sent to a senior editor at the journal November 21st in which he forwarded the exchange with Maucuer and mentioned sending her the evidence of data manipulation and retracting the paper. 

She replied within a few hours: 

Many thanks for your emails. I am sorry to hear about the unfortunate incident involved in your published paper. Please bear with us while I discuss these matters with my colleagues and my Chief Editor. I will respond to you as soon as I can.

In Tsichlis’ reply later that day, he said he hoped the paper would be retracted “as soon as possible.” He also expressed a hope that “any additional publicity will be avoided, because it does not really serve a purpose.”

Over the next two months, Tsichlis and the editor exchanged a few emails, but it wasn’t until January 18th that the editor, apologizing “for the extreme delay” sent a retraction form with proposed text for the notice: 

The authors are retracting this Article as an independent researcher brought to their attention data irregularities in the reporting of U2AF2 exon 2 alternative splicing. On further investigation, the authors cannot confirm that the proposed splicing mechanism will give rise to a functional U2AF2 protein. The authors cannot confirm the integrity of the electropherogram in Figure 1g. In light of these errors, the authors have no confidence in the key findings of the paper, and therefore, wish to retract the Article.

Tsichlis responded on January 26th, saying that that the statement was fine, but he’d like to “make it a little more specific.” He proposed this version: 

The authors are retracting this Article as an independent researcher brought to their attention the fact that his bioinformatic analyses of multiple data sets could not confirm the alternative splicing of the U2AF2 exon 2. On further investigation, the authors observed that the proposed splicing mechanism could not give rise to a functional U2AF2 protein. In addition, the authors identified errors in the electropherogram in Figure 1g, raising questions on its integrity. In light of these errors, the authors have no confidence in the key findings of the paper, and therefore, wish to retract it.

The next day, a staffer from the OSU research integrity office who had been copied on the emails suggested changing “these errors” to “this information,” “with the reasoning being that ‘errors’ has a specific meaning that we may not want to convey.”

The following Monday, January 31st, the editor thanked Tsichlis and the staffer for the edits and asked them to update the retraction form she’d sent earlier and sign it so she could discuss the changes with her colleagues. Tsichlis did so later that day. 

On Friday, February 4th, the editor wrote back that after discussion with her colleagues, the editorial team “felt that the initial retraction text is more appropriate.” She asked Tsichlis to reach out to his coauthors to sign the form. 

Tsichlis responded on February 7th: 

Thank you for your answer to my earlier Email. The fact is that the two statements are not different, except the statement I wrote is more specific, which I believe should be what we want. The truth is that I would like to retract this paper because what it says is wrong and I am very embarrassed by it. At the same time, I would like to describe the events that led to the retraction, as clearly as it can be. First, I want to be transparent by pointing out that indeed the process was initiated with the Email I received from an independent investigator. However, the description of the events, as they happened after this Email, should also be truthful and clear, with no ambiguities, so that anybody who reads it understands what happened. This is important to me and all the other people who had the misfortune to be involved in this work. Please, find in the attachment the statement you wrote with my comments, the statement I had written, and a new statement I wrote now, which is even more specific than my earlier statement. Unless you think that what I suggest is factually incorrect, I am not sure why it cannot be used.

Thank you and I do apologize for my persistence on this.

His new version of the statement spelled out the issues he’d discovered without euphemisms: 

The authors are retracting this Article as an independent researcher brought to their attention the fact that his bioinformatic analyses of multiple data sets could not confirm the alternative splicing of the U2AF2 exon 2. On further investigation, the authors observed that the proposed splicing mechanism could not give rise to a functional U2AF2 protein. In addition, the authors confirmed that the electropherogram in Figure 1g, was generated by artificial splicing of two separate electropherograms. In light of this information, the authors have no confidence in the key findings of the paper, and therefore, wish to retract it.

That email is the last one in the records we received from OSU on May 6, after requesting them on February 2. We reached out to the Nature Communications editor to ask about the status of the paper, but she did not respond. A spokesperson for the journal sent this comment: 

For confidentiality reasons, we are unable to comment on individual cases. However, it may be helpful to note that in general, when concerns are raised about papers we have published, whether by the original authors or by other researchers and readers, we look into them carefully on a case by case basis, following an established process, including consulting with the authors of the paper and, where appropriate, seeking advice from peer reviewers and other external experts. More information about our corrections and retractions policy is available on our website: https://www.nature.com/authors/policies/corrections.html.

Laliotis didn’t respond to our request for comment. When we reached out to Tsichlis, he didn’t want to discuss the experience, citing an ongoing investigation at OSU. He did send us his slides documenting the data manipulation he found, along with this comment: 

I do not know why “Nature Communications” has not retracted the paper yet, although I requested it and I provided all the evidence they should need to do that (see attachment). Their sister journal, “Communications Biology”, retracted a paper that was based on the same original finding at the end of 2021. I will contact them again to see why they procrastinate. 

The paper has been cited four times. Three of those citations came after Tsichlis requested its retraction last November.

Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].