Longtime readers of Retraction Watch may recall a 2011 post about a research team that retracted a paper after realizing that they had ordered the wrong mice. Maureen Gannon and Raymond Pasek of Vanderbilt University contacted us earlier this week to alert us to a similar case: Their retraction, earlier this month, of a 2016 paper from American Journal of Physiology – Endocrinology and Metabolism after discovering that “a colleague from another lab had mistakenly supplied us with the wrong transgenic mouse line.”
“We strongly believe that sharing this example will encourage other researchers to do the right thing when a mistake is discovered and promote academic integrity,” they wrote. So we asked them to answer a few questions about their experience with “Connective tissue growth factor is critical for proper β-cell function and pregnancy-induced β-cell hyperplasia in adult mice,” a paper that has been cited twice, according to Clarivate Analytics’ Web of Science.
Retraction Watch: How, and when, did you become aware of the error?
We first became aware that something was wrong with the Cre line we were using when additional studies (following the publication) revealed impaired glucose tolerance in Cre-only animals with no additional genetic manipulation. This had not been observed in the original cohort of Cre-only animals that were used as the basis of the AJP paper. However, on certain genetic backgrounds the Cre line we were unknowingly using (RIP-Cre) is known to have impaired glucose tolerance and reduced insulin secretion. Thus, the phenotype likely developed over time with years of breeding. Our first indications of this phenotype were in November 2016. At first we thought that the Pax6-Cre (which we thought we were using) had unreported phenotypes in glucose homeostasis. However, as time went on, we became convinced that we were not in fact working with the Pax6-Cre line as we had been told. In early 2017, we began to suspect that the mice were in fact the RIP-Cre line known to have issues with glucose homeostasis. We were able to confirm this suspicion using RIP-Cre-specific PCR primers in April 2017.
RW: What was your reaction when you realized these were the wrong mice?
To put it simply, we were (and still are) devastated. To be honest, this is my (M.G.) worst nightmare as a PI. I have always told everyone in my lab that I always want the truth and that I don’t care if their results contradict or disprove a hypothesis. Having to retract a paper is extremely upsetting to me and to the first author (R.C.P.). This study represented years of work and resources and we were proud of and excited about the paper. We immediately told our colleague who had given us this mice, concerned that there was incorrect genotyping in their mouse colony and that they too were mistakenly using the RIP-Cre mice for their studies. We then contacted the journal editors to let them know and ask them how to proceed.
RW: Did you have any concerns about coming forward, given the stigma of retractions?
Of course. It is embarrassing to us to have this come forward, despite their being absolutely no fault on our part that this happened. Most people trust the veracity of a published reagent when given by a colleague. However, we knew that it was the right thing to do to alert our colleague and the journal to the mistake as soon as we realized it. We did not even hesitate about making this decision. I always want the science coming from our lab to be trusted and reproducible. Ignoring this, or sweeping it under the rug would undermine our credibility.
In addition to alerting our colleague and the journal, I immediately informed my division chief, my department chair and the dean of faculty affairs. These were not fun conversations. But I knew that it was the right thing to do. As my chair told me “You are not defined by things that happen to you that are beyond your control. You are defined by how you handle these situations.”
RW: When did you alert the journal to the problem? What was their reaction?
The same day we obtained the genotyping results showing us that we had indeed been given the RIP-Cre mice instead of the Pax6-Cre. Luckily, we had saved the DNA from the mouse we were given by our colleague several years ago and could compare with known mice of each transgenic line.The editors of the journal could not have been nicer. They completed an investigation and found no misconduct on our part. They felt very badly that we were going through this and repeatedly told us how much they appreciated that we had come forward. They helped us write up the retraction notice that was published in the July 2017 issue of the journal.
RW: Were you aware of other cases like this, such as this one from 2011, or the literature on retractions for honest error? If so, did that guide your approach here?
We were not aware of the case that you alerted us to. However, we have read about many retraction cases posted on your site and have read about cases in which there was no misconduct or fraud and authors voluntarily retracted a paper due to honest error.
RW: How have your colleagues responded to the retraction?
We have received nothing but sympathy and support from everyone we have told. They all agree that we did the right thing.
RW: What advice would you have for other researchers who realize they’ve made an error that has a dramatic effect on a paper’s conclusions?
Honesty is always the best policy…even if it’s hard. Especially in today’s climate of “fake news” and science being under attack. The scientific community and the public needs to trust published scientific results. We should be able to admit when a mistake is made and correct the literature. This has been a learning experience for us and an opportunity for growth.
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 daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.
Bravo!
This case reads almost like a distraction from the phenomenon of “C57BL6” mouse, wherein thousands of authors have no idea whether they are using C57BL/6J, which has “issues” with glucose homeostasis, or C57BL/6N, which has not.
DOI: 10.1186/1745-6150-9-18
Another example is the αMHC-MerCreMer transgene, which is featured in about 200 publications, lately found to have a large transgene-derived genomic deletion
DOI:10.1007/s11248-016-9960-6
This is still a problem in the beta-cell field, but in this case the glucose intolerance is due to unanticipated expression of full-length processed and active human growth hormone from the hGH minigene included in the original RIP-Cre targeting cassette (and many other Cre lines), increasing mouse beta-cell proliferation and thus decreasing their ability to function.
https://www.ncbi.nlm.nih.gov/pubmed/25470546
If everyone used the right LITTERMATE controls, none of this would be a problem. Test Cre only littermates and LoxP only littermates, and you’ll see any problems immediately. Oh, and genotype every pup, regardless of the breeding strategy.
Hi Dave,
Thanks so much for the feedback. It’s great to hear from other member of the scientific community on our work. We absolutely agree that proper controls are critical for any experiment, but that was not the issue here.
As the blog post (and our original manuscript) indicated, our studies did examine Cre only controls. These animals had identical glucose tolerance when compared to LoxP only controls. In fact, even a rather costly glucose stimulated insulin secretion assay did not reveal a phenotype when samples with the Cre transgene were compared to LoxP only samples. The Cre only phenotype did not start emerging until later after multiple breedings. Unfortunately for us, it’s impossible to determine exactly when. In fact, even as of this year the Cre-only phenotype was not present in every mouse examined.
To clarify, the RIP-Cre transgenic mouse line does not always display a phenotype on its own, it only manifests under certain genetic backgrounds. The issue here wasn’t that we didn’t use the proper controls, it’s that we weren’t actually given Pax6-Cre transgenic mice.
We certainly hope this unfortunate situation will promote more positive discussions about honesty and scientific integrity. If you have any other concerns or comments, please feel free to leave your name and contact information for us. We’d love to hear more suggestions about how events like this can be prevented in the future.
Cheers,
Raymond Pasek (first author)
So if I understand this correctly, the Cre+ line only displayed a phenotype much later, and after you had already published the AJP paper with earlier generations of Cre+ mice as controls? The issues with the control could happen for a number of reasons that might not be anything to do with the Cre transgene per se. The most likely would be genetic drift caused by, for example, repeated Cre+ x Cre+ breeding. If this is the case, then your original data could still be OK if you really did use Cre+ controls in that study. Unless I am missing something? Have you obtained fresh RIP-Cre from JAX and compared them to your RIP-Cre?
There are no “issues” if you use the right controls.
Can we go back to this? I’m really confused as to how this is due to RIP-Cre and not genetic drift if the correct controls were used. And, if the authors used littermate controls for each study, that would cancel out any results due to genetic drift since all of the mice should, in theory, have the same mutation, right? I really want to understand what went wrong here. I’m implementing SOPs for breeding/line maintenance/what is a proper control in the lab and these examples are really great ways to learn. Thanks!
A very important example and a model of honesty and integrity in science. To be good at science mean to be honest above all!! BRAVI again.
“If ambivalence is one of your issues [in making a decision], you probably do not have a well-defined [moral] code” -Max Strom. I’m sure this was an agonizing decision, but it was done with grace and humility and the scientific community is better off because of it. I hope I never have to make the same decision, but I know what that decision would be.
Were generic cre primers used to genotype the mice, and only specific cre primers used later?
It’s good that they came forward, but as a mouse geneticist I have to say they were extremely careless in how they performed their experiments. Standard practice when receiving a new mouse line (especially something from another academic lab) would be to genotype with gene specific primers (i.e. not a general Cre PCR). Also Cre lines should be crossed to a reporter (i.e R26R) to ensure that they give deletion in the correct tissue pattern. You should NEVER EVER just take a mouse and start doing experiments. In fact reading through their retracted paper, nowhere do they actually demonstrate that the deleter line they are using is actually knocking out the CTGF gene!
Hi Inigo,
Thanks for your comment. For the record, we did confirm Ctgf deletion in our tissue of interest. Both the Pax6-Cre and the RIP-Cre induce gene deletion in the beta-cells, so this method did not alert us to anything being amiss. We simply didn’t include this data in the publication. The mouse line we thought we were using was first published in 2000, so publishing data that shows the Cre works wouldn’t be very useful and relevant.
Additionally, as the retraction notice states, Pax6-Cre specific primers are not currently available. This is sadly due to the fact that the sequence to the Pax6-Cre transgene is not published. We reached out to multiple labs that have used the Pax6-Cre mice in the past and we were unable to obtain the sequence and design Pax6-Cre specific primers, thus we used generic Cre primers. The mouse line we were given expressed Cre in the cells we expected it to, induced deletion of our LoxP gene, it genotyped positive for the Cre, and it (at the time) did not show any phenotype on its own, so we had no reason to suspect anything was amiss.
If you have any other concerns or comments, please feel free to leave your name and contact information for us, we’d be happy to get more feedback on this.
Cheers,
Raymond Pasek
“we have read about many retraction cases posted on your site and have read about cases in which there was no misconduct or fraud and authors voluntarily retracted a paper due to honest error.”
There are also many cases where there is research misconduct. Good to see this is not one of them, well done Ray.
Minimum standards for breeding mutant mice: http://onlinelibrary.wiley.com/doi/10.1111/j.1601-183X.2008.00438.x/abstract
Knowing Dr Gannon, this doesn’t surprise me. I cannot applaud her and her lab’s honesty and courage enough.
I’m confused by this story. Wouldn’t the entire issue been avoided by using the proper controls? If the difference really was due to RIP-Cre, had the authors used Cre(+)GOI(+/+) and Cre(+)GOI(d/d), a difference would not have been detected, correct?
Unfortunately, no. Both Cre lines express in the beta-cells, so the gene of interest is deleted in either case. During our work, we did confirm deletion.
Raymond Pasek