Misidentified DNA leads authors to retract zebrafish cholesterol paper


Authors are retracting a 2012 paper on cholesterol metabolism in zebrafish after realizing it included a case of mistaken identity in a DNA sequence crucial to some aspects of the experiment.  

A postdoc misidentified the plasmid in question after failing to fully sequence it before including it in the experiment. A technician in the lab found the mistake, last author Steven Farber, a researcher at the Carnegie Institution for Science in Maryland, explained:

When the omitted region was correctly sequenced we discovered it had an error.

He told us in a phone interview what that felt like:

We were like, holy crap.

Next came months of back and forth with the journal, discussing whether to correct or retract the paper. Farber tells us the mistake, which affects two figures,

was unfortunate. Most of the paper is in fact correct.

The paper, “Visualization of Lipid Metabolism in the Zebrafish Intestine Reveals a Relationship between NPC1L1-Mediated Cholesterol Uptake and Dietary Fatty Acid,” published in Chemistry & Biology, has been cited 21 times, according to Thomson Scientific’s Web of Knowledge.

Here’s the detailed retraction note

This article has been retracted at the request of the authors.

In our original Resource, we reported a regulatory link between NPC1L1-mediated cholesterol uptake and dietary fatty acid in zebrafish larvae. In the absence of dietary long-chain fatty acids (LCFA), larval enterocytes failed to internalize luminal BODIPY-cholesterol. We attributed this LCFA dependence to the cholesterol transport protein NPC1L1, as dietary LCFAs were shown to induce the translocation of transgenic human NPC1L1 from a perinuclear compartment to the intestinal brush border (supported by the data shown in Figure 5 of the Resource). When overexpressed, human NPC1L1 localized to the brush border, where it was able to mediate the internalization of BODIPY-cholesterol even in the absence of dietary LCFAs (Figure 5). We also reported that NPC1L1 directly mediates cholesterol uptake by larval enterocytes (shown in Figure 6 of the Resource). However, we recently discovered that the larval expression of the human NPC1L1 transgenic construct hsp70:HsNPC1L1-mCherry we reported was in fact that of hsp70:mCherryCAAX, which encodes an mCherry fluorophore modified with a prenylation motif. When employing a recloned hsp70:HsNPC1L1-mCherry vector, our experiments failed to replicate the LCFA-induced translocation of NPC1L1 to the intestinal brush border, indicating that the findings reported in Figures 5 and 6 are no longer valid.

Our conclusions regarding zebrafish intestinal lipid droplet formation and depletion when given a high-fat meal, BODIPY-fatty acid incorporation into LDs, and BODIPY-cholesterol localization to the endocytic compartment of enterocytes, distinct from LDs, are not affected by this mistake. Additionally, the finding that dietary cholesterol absorption is dependent on luminal long-chain fatty acids is also unaffected. All of these experiments (Figures 1–4 in the Resource) were performed using zebrafish lines that did not contain the incorrect construct mentioned above.

Nonetheless, given that some of the core mechanistic conclusions we presented are no longer valid due to the use of the wrong construct, we are now retracting the paper. We apologize to the community if our error caused any significant confusion or inconvenience.

Farber told us that the process of checking the work, and working out what to do, took months.

It’s not easy — you don’t want to retract it if you’re not 100% sure it’s not right. We did new experiment as quick as we could. I put a new figure together. The editor said, ‘I want you to correct [the paper].’ [The correction] was going to proofs. It went out to review to the original reviewers. And then they decided, ‘you know, in retrospect, we think we’re going to retract it.’ We wasted a few months. I still wanted to know the truth anyway.

He added that he wasn’t satisfied with the way that the current publishing system handled the mistake:

I think the larger issue is you print something and then it’s solid for ever. It would be amazing if there was a way we could publish something and then build to that and add on it. I’ve been thinking about how to do this, it’s an interesting challenge — trying to keep the publication live. We’re stuck in this old method, where it’s locked in time. So much of [the retracted paper] is right.

Farber noted in an email that they were re-doing experiments that did not involve the plasmid, and would publish them in other papers:

As far as future work, we are planning on republishing portions of these findings (new experiments done by different investigators) in newer projects but I [have] no plans for using any of the actual data used in that paper.

We asked him how the problem could have been prevented:

Sadly, about a year after the construction of this plasmid I put in place a system whereby all plasmids from lab members are partially resequenced by my technician prior to going into our plasmid database.  This would have caught this error.  Secondly, all future plasmids in Farber lab publications will be sent to Addgene (https://www.addgene.org/) which is a nonprofit organization that can accept plasmids for free and distributes them for a small fee to any lab worldwide.  They independently sequence submitted plasmids — that too could have caught this mistake.

We’ve reached out to the journal’s general email address, and a spokesperson for publisher Cell Press got back to us:

We feel the retraction notice is clear on why the authors thought it would be best to retract the paper. The editors support the authors’ decision.

We asked about the initial plan to correct the paper. The spokesperson confirmed that was true, and explained that

after multiple discussions, the authors concluded that retraction was the most appropriate option based on advice from the editors.

Farber was the last author on another retraction in PNAS we reported in 2013, after the first author was found guilty of scientific misconduct.

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, and sign up on our homepage for an email every time there’s a new post. Click here to review our Comments Policy.

One thought on “Misidentified DNA leads authors to retract zebrafish cholesterol paper”

  1. It’s good to hear that the authors have resolved the problems with the journal.

    Having said that, I don’t quite agree with one of the author’s comment about the current publication model. In particular, the statement about “we could publish something and then build to that and add on it”. Yes, it is a bit unfortunate, but that is why a publication is so coveted by everyone from graduate students to PIs. It’s because it is hard to do and there is a final point in time where it is done and you cannot go back. So, you better make sure it is correct before that point in time.

    As a counter-example, there are situations where you can “publish” something and then go back and fix it. With computers, there is computer software and software patches. Software companies have made us accept that software and security patches are a necessary part of the software lifecycle, but there was a time when software came on disks/discs and before you ship your software, it had to be perfect.

    I understand what the author is saying…

    But, I would hate to reach a point in scientific publishing where if it’s published in January 2016, I read it, and build on that work and then find out in June 2016 that it’s been corrected and what I had been doing for 6 months was wrong. I need to continually check all the papers I’ve read to see if minor updates have been done to it…

    Somehow the possibility of a retraction will make authors (including me) more serious about getting it right the first time?

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.