Cell runs a lengthy correction, rather than retraction, for image problems

The journal Cell has an interesting — and somewhat puzzling — correction this month that we’ll add to our “mega-correction” file.

At issue is a paper, published in October, from the lab of Harvard’s Stephen Elledge, a noted genetics researcher, whose first author is a post-doc there named Michael Emanuele.

According to the notice, Emanuele (singled out, we note) seems to have been rather careless with the images used in the article, titled “Global Identification of Modular Cullin-RING Ligase Substrates”:

Since publication of the above article, it has come to our attention that incorrect immunoblot loading controls in Figures 6 and S5 were selected by the lead author, M.J.E., during figure preparation. Corrected versions of the figures appear below and are available with the article online. In each case, the relationship between the experimental and control data is clear on the original films. We apologize for these mistakes and emphasize that their correction does not affect any data interpretation.

The notice then shows the figures, followed by more text:

In Figure 6C, vinculin controls were mistakenly taken from the experimental rather than the untreated controls. The full experiment can be seen in a longer exposure in Figure 7D. The correct loading control is now shown in the revised version of Figure 6.

In Figure 6E, panels for PAF15, cyclin B, and phospho-S10 on histone H3, showing the cell-cycle position for each of the extracts, were previously published in Emanuele et al., 2011, and should have been cited as such. These immunoblots and corresponding quantification have since been repeated, and the new blots and graph are shown in the revised Figure 6. The original quantification shown in Figure 6E incorrectly normalized the PAF15 and cyclin B data from the earlier analysis to the vinculin control performed for this study.

In Figure 6F, cells were treated with siRNA targeting FF or cyclin F. The loading control shown for siCyclinF was incorrectly duplicated from the siFF panel and is replaced with the correct panel here. The original quantification was derived from the correct loading control.

In Figure S5D, the vinculin control for the 293T cells was inadvertently duplicated from the U2OS cell loading control. The correct control appears here.

The panels that “were previously published in Emanuele et al” appeared in 2011 in PNAS (Emanuele et al., 2011 Emanuele, M.J., Ciccia, A., Elia, A.E.H., and Elledge, S.J. (2011). Proliferating cell nuclear antigen (PCNA)-associated KIAA0101/PAF15 protein is a cell cycle-regulated anaphase-promoting complex/cyclosome substrate. Proc. Natl. Acad. Sci. USA 108, 9845–9850.)

Emanuele, whose name appears on 11 articles, according to a Medline search, is still affiliated with the Elledge lab, where we tried unsuccessfully to reach him. We’re also trying to find out more from Elledge himself, and will update with anything we hear.

The blog Abnormal Science wrote about this paper (read both links) in December, saying that it learned about the suspect images from a “vigilant whistleblower.” The study has been cited twice, according to Thomson Scientific’s Web of Knowledge.

As readers of this blog are doubtless aware, recycling Western blots helped earn Naoki Mori a 10-year publishing ban from the American Society for Microbiology, along with a lengthy suspension from his institution. It also led to more than 30 retractions.

We’re curious why the two cases had such different endings. We know this much: The variable isn’t the publisher. Cell is an Elsevier title, but Elsevier also runs journals that published Mori’s work, including Biochemical and Biophysical Research Communications and Leukemia Research.

We asked Cell, through an Elsevier executive, to comment on the apparent discrepancy and will update this post when we hear back.

Updated 5:20 p.m. Eastern, 2/2/12:

We spoke with Elledge, who insisted that the case was a matter of disorganization, not misconduct.

Everyone who has seen the primary data agrees that this isn’t a case of data manufacturing, this is a case of errors in the assembly process of the figures.

Elledge said he learned of the problematic figures from Abnormal Science in early December, upon which immediately took possession of Emanuele’s data and notified Cell.

 There’s a really bad feeling that comes on. You don’t know what to think. It sure looked bad.

But Elledge said he quickly realized that his post-doc’s records were well-kept and thorough, and that the problem stemmed from failing to double-check the figures.

 Fortunately this was all on film; this wasn’t electronic. You cannot manipulate the films, you’d have to have done something way in advance.

In the end, said Elledge, a correction, not a retraction, was the appropriate step.

There were multiple mistakes, which makes it look bad, but this is a case where there were just mistakes.

Hat tip: Pinko Punko

57 thoughts on “Cell runs a lengthy correction, rather than retraction, for image problems”

  1. “We’re curious why the two cases had such different endings”

    I suspect this has something to do with who we are talking about here (HHMI investigator etc). This guy and his lab are churning out 5 – 10 CNS papers a year. I’m not in this field and had no idea who he was, but I am genuinely astonished by the productivity. Hard to know what to make of this story though…..

    1. What a mess. It would be very interesting to see the “notebooks” in the Elledge lab. I’m guessing the typical 3 ring binder disaster.

  2. He is a top notch scientist and does interesting and excellent work. If we take him at his word, and I do- in his response he goes further than the correction and states that they looked at the raw data, which they had in hand (good), that the original data were intact and supported the conclusions (good). The issue here is with a paper with so many authors, the only way to validate the data for the figures when 1 author is making them is to basically say: “show me the raw data from the notebooks that are in these figures”- this is what should be done basically. The guy should have little tabs in his notebooks that say “Fig. 4B loading control” or whatever- I think that is the best way to do it. The question may be what is it about the way the first author put together his figures and annotates his files that make mistakes happen where blots show up in more than one paper. We all make mistakes, so to some extent I support the “mega-correction” within reason. It is a tough position to be in.

  3. Pinko Punko: Srsly? Top notch scientists do a better job running their labs. It’s all about the culture of “zero tolerance” you establish in your team. This single incident and the solid defense of someone who at best is a sloppy scientist tells me he’s no top notch scientist – certainly no role model in my book.

    1. I’ve followed Dr. Elledge’s work for a long time. For the most part his work is beyond reproach and he has been a significant contributer to a number of fields.

      That said, Fig. 6 (even in its corrected form) is hardly representative of the work of an excellent scientist. I suspect the reason is that Elledge had little to do with the production of this paper other than financially supporting a few of the authors and lending the prestige of his name to an otherwise average publication.

      1. Yeh I agree with you MM and I think this is an incident where the Retraction Watch blog needs to be VERY careful. Having spent some time reading about this guy, this is clearly a scientist who has earned some serious respect and I am quite certain that the post-doc in question here has been severely reprimanded. I do not think we will see the typical domino effect here. The only thing you can accuse Elledge of is not spotting the schoolboy error by his staff but, looking at the size of his lab, it must be a very difficult task indeed.

        Now, whether mega-corrections should be allowed as opposed to a flat out retraction is a different debate, really.

    2. Well, you would have to know the fields (plural) that this guy has contributed to and the huge impact he has had. That said, a paper with that many authors should not have duplicated western blots. The other authors should have done a better job. Obviously as corresponding author, Steve Elledge has the responsibility, and signs paperwork to that extent. In one way the correction is unfair in singling out the first author, in other ways it is important to illuminate and be transparent with what happened, so I think that this overall was the correct approach. Nobody wants to be in this position.

  4. please. what are the odds that ONLY the vinculin blots are the problem. one of the new vinculin blots shows very uneven loading. ok fine you are a slob in your notebook, but the data already published in PNAS??? you repeat that ??? how is this not fraud. elledge should be ashamed of himself to stoop so low.

    hey steve: have some balls and retract it. wimp. you screwed up so admit it you are the pi. you published the same figure in two papers!!! f this paper leads to the nobel YOU will get it and not that crappy postdoc who did the work.

    what is one less cell paper gonna do to elledge’s career??? zilch. he will be hhmi for life, he doesnt even need the nih. i am SURE being a cell EB member has NOTHING to do with this. these guys are the 1%’ers. we are the 99%. the first author will get the dream job in SF, SD, palo alto, or boston.

    1. “these guys are the 1%’ers. we are the 99%. the first author will get the dream job in SF, SD, palo alto, or boston.”

      I chuckled!!!!

    2. agree with cugel. It is amazing truth that only people who publish in Nature, Science and Cell get brilliant ideas and usually we trust those papers. Pinko Punko has spotted the correction, I believe and please have look at the Abnormal science blog link above. How come the same mistake happened in more than two papers. A domino effect too many? Having said that, there are cases where science was good and profile of the scientist was good, still things like this happen.

      1. It means people who are publishing in small journals are stupids? This is a similar issue like that of Aggarwal. Why HHMI or Harvard is not investigating this? same Elledge lab was spotted earlier in PNAS publication and journal asked for original gels. I didn’t see any RW posting on that? I think there is some thing fishy going on to target people of non-US origin,

      2. @Aktfamily: not sure about the targeting particular category of people. both have influence in the field. Implications may have huge consequences financially to the institutions!! My guess is that nothing is going to change in both the institutes. All ends well – people have short memories any way.

    3. I was prepared to argue that the an overworked and disorganized postdoc could accidentally switch data that was being done simultaneously for two papers (the Cell and PNAS papers). Then Dr. Elledge commented that “he quickly realized that his post-doc’s records were well-kept and thorough”. Given that, I don’t know how one would accidentally combine results from two separate experiments.

      1. You are assuming it was done by accident. My opinion is that the post-doc genuinely thinks it is OK to re-use the same loading control, and that is not a good reflection of the lab as a whole. I mean, lets face it, running loading controls when you know the outcome of the blot is a little on the dull side, so I think that some just don’t see anything wrong with re-cycling the same blot if the experiment is essentially the same. However, when you are doing siRNA treatments etc, it is not simply a “loading control” anymore; it is an essential control to show stability of a reference protein. Your membrane may look perfect on the ponceau stain, but there could still be off-target effects of your treatment and it is imperative to show a unique control for each experiment, especially when you are going to present quantitative analysis. This seems to have been lost on this post-doc. The worst part is he is using the same control blot when experiments are done on different CELLS!!! That is not good at all and should have been caught.

  5. There are 12 authors on the Cell paper! I cannot comprehend how this can happen when all 12 authors “read and approved the work” prior to publication. Furthermore, all these authors must certainly have been aware of the PNAS paper published just months before by the same lab, with some of the exact same data. Where were the reviewers? Perhaps this lack of scrutiny and cavalier approach to publishing is one of the many perks of the 1%?

    This is a classic example of what happens when labs get too large, and the PI is presumably too busy to notice such obvious irregularities. It is frightening to realize the lack of scrutiny that, by definition, accompanies the research produced by these top labs. Maybe it is time we realize that it isn’t humanly possible to consistently perform thorough, accurate, and high quality research in a mega-lab.

    This incident reminds us that we must be even more skeptical of the research published by the top 1%

      1. about to write about the two corresponding author. The other one is listed as a postdoctoral fellow on the lab webpage..big boost for the junior colleague…

      2. I am corresponding author on all publications from my PhD study, as well as on one paper from my Master’s project; it’s a matter of what the PI of the lab considers appropriate.

    1. Maybe the same thing happened with Imanishi-Kari & Baltimore, and Van Parijs & Baltimore. When a lab gets too big, and the PI does not carefully read the papers with his name on, standards fall. While a solution might be for these PIs not to accept honorary authorship, without their names the papers would not even get reviewed by the big journals.

      1. This has precisely been my point when people defend the PI on the basis of their workload and the number of people working under them. When they take on a graduate student or a post doc, it should not be getting another slave to work in the lab, produce data and publication on which the PI can get the name. It is mean to be a commitment to supervise their research and mentor the career path of the students and post docs. Why are funding agencies silent on the lab size when now mentoring and training has been made a part of grant evaluation process?

      2. @expostdoc:

        The NIH, and similar organizations, may have started to play lip-service to “mentoring and training,” but what would you actually have them do? For one thing, how would you evaluate effective mentoring? Output? Hours spent with lab members? Career placement? None of these things would be an effective measure. Mentoring is very personal; people arrive in a lab with different strengths and weaknesses. Some people work very well in a large lab with a hands-off environment, while some people flounder.

        The Ph.D. is, effectively, an apprenticeship, as is most postdoctoral work. Students and postdocs have a responsibility to do some research on labs, and to evaluate whether the lab they join will be an effective training environment. Large labs don’t hide what they are, and the current lab members do provide a sense of the working environment.

        The choice of apprenticeship is largely up to the student/postdoc, and it’s *not binding*. If you don’t like what you chose, you can walk away. The blame for these factory-line labs does not fall solely on the PI. The students and postdocs have joined the lab for a reason, and can choose to look for other labs if the situation isn’t working out.

        (I realize that the situation can be somewhat different for foreign students, but I think addressing this as a visa problem would be more effective than trying to force a personality change on a PI).

      3. @Sfs
        You yourself said … paying lip service .. that is what it is.
        It is absolutely true that it is upto the student / post doc to choose or not to choose a lab. At the same time it is also not binding on the PI to take on people if he/ she thinks that the work load is too much for them and they will not be in a position to directly supervise. Isn’t their name and reputation at stake? Why should they be allowed to get away with the excuse of being overburdened with other responsibilities? I am not saying that they need to monitor every little thing that is being done in the lab .. that is humanely impossible .. if a student/ post doc wants to fabricate data they will find a way.
        As for mentoring part … is it really possible for the PI to think about the strengths and weaknesses of 30-40 people working under them? Lets be a little honest over here … basically you either have the critical aptitude or you don’t have it to make it science. You acquire technical skills during the phd .. some more during post doc, if the PI allows you can learn some thing about writing and presentation. Period.
        Over the last couple of decades the time being spent as post doc has increased. Why? Is it that ‘mentoring’ nowdays during PhD isn’t of the same quality and hence they need to be mentored for longer time? No .. earlier you could be an independent faculty immediately after PhD or may be a couple of additional years but now less than 5 yrs .. very rare! NIH has fixed 5 yrs as max but still people hang on upto 7 yrs till they find a position. Its just a question of how you spend your time till you find a position, work with some one else’s funds till you get your own funding, add some more publications to your CV. This is purely a question of demand and supply … despite the huge gap import is still encouraged as that fuels the PIs research at low cost.
        Mentored research is the term used but its just more experience since science is a continuous learning experience. The more years you spend the better you get. It doesn’t stop once you are in faculty position. The depth of knowledge, the authority in the subject grows. Technology gets outdated so quickly, new techniques being developed and if you want to keep up you keep learning.

    2. I can tell you from my personal experience as to how these kind of things happen. In my previous lab there were several co-authors on the papers but the all the authors never got to see the paper. The paper was written exclusively by the first author and the PI. Some times some of the coauthors got to see and give their comments on the paper but again not all of the co-authors. As for signing the copyright form there was one instance where the authors did not sign … not because of any conflict but they were not there … I was asked by my PI to sign as the first author .. another student was called to sign as one the co-author, a post doc who was no longer in the lab .. and so on. Different people were asked to sign so that the signs would not look similar and job done.
      One of the post doc who was a co-author on one of the grad students paper would ask me the status of that manuscript as I was friends with that student so that accordingly he could modify the publications on his CV without having to ask the PI or that student as he did not like him.
      So its only on paper that all authors contribute, read and approve .. reality is different. This works for the PI as that way they can have full control over authorship .. include anyone and say contributed or exclude. I had to deal with that too .. despite all evidence of contribution to the project I could still not take it up as research misconduct as I did not fulfill 2 of the 3 criteria – scholarship and approval of the manuscript .. they would not answer as to how one can have scholarly contribution or approval part if the PI though using data from my experiments would not give the opportunity. Another well known name, another top institute …

      1. Co-authors do not benefit from their co-authorship on Cell, Nature and Science papers ? or other papers….? please do not be so niave….

        A co-author shares equal responsibility, otherwise why is he a co-author ?

      2. @scienceobserver:
        You clearly do not understand scientific authorship. Credit is not shared equally between authors, and neither is responsibility. There is a reason why fights over first authorship can be so vicious.

        In biology, chemistry, and some areas of physics, the authorship is quite hierarchical. In terms of credit, this is the order:

        1. Corresponding author(s): funded the project, designed the project, (hopefully) substantially edited or wrote the manuscript, and (also, hopefully) guided/oversaw the entire project from start-to-finish

        2. First author(s): did the bulk of the work on the project, substantially contributed to its design, and (often) wrote the bulk of the manuscript

        3. Everyone else: they did something, but it’s often not clear how much (or what) they did. Assuming that all authors made a legitimate contribution, this contribution may have been a small but important experiment, unusual reagents, a statistical analysis, etc.

        The first and corresponding authors gain the most from the publication. By “most,” I mean 90% or more of the credit, not 51% of the credit.

        Put another way: In academics, only first-author papers truly count toward a permanent position. Second-, third-, etc.-author papers are nice, but even 100 second-authorships does not equal a first authorship. Similarly, only corresponding authorships really count for tenure decisions.

        So, yes, the first and corresponding authors bear the responsibility for most errors in the manuscript. They had control of the manuscript, and they received almost all of the credit for it. Even if it’s a Science, Nature, or Cell paper.

    3. @Data Skeptic:
      “There are 12 authors on the Cell paper! I cannot comprehend how this can happen when all 12 authors “read and approved the work” prior to publication.”

      Let’s be realistic. All twelve authors did not carefully read the entire manuscript, or obsess over the figure. All of the non-first, non-corresponding authors probably read/skimmed the manuscript, double-checked their own contribution, and assumed that the first and corresponding authors have quintuple-checked everything.

      The first and corresponding authors receive the vast majority of the credit for the work, and shepherding the manuscript is their responsibility. In a lot of collaborative studies, not all of the authors have the expertise to check the rest of the figures. Even when they do, it’s a matter of balancing time commitments: does the author spend the time vetting a manuscript which will probably not bring them closer to a permanent position, or do they check their own contribution and then continue working on projects that will have a better payoff?

      Maybe this should be different, however you’re fighting with human nature if you want the secondary authors to spend a chunk of extra time doing something that provides no clear reward.

      1. In a lot of big US labs there are many more post-docs (mostly foreign) than PhD students. The post-docs do most of the experiments, and if they are experienced enough, write and submit the manuscript. The PI’s role is to get funding, to influence editors, and to present the results at meetings. None of this requires reading the paper or looking at the original data.

      2. @Michael Briggs:
        As a former member of a medium-to-big US lab, I am aware of how things happen. It’s a huge problem when PIs are so hands-off that they don’t look at the original data. There should certainly be consequences.

        However, the original comment wasn’t really about the first or the corresponding author: ‘I cannot comprehend how this can happen when all 12 authors “read and approved the work” prior to publication”‘. This, in effect, spreads the blame from those who control the manuscript (the first and corresponding authors) to those who don’t, and who gain very little from carefully vetting a manuscript they don’t control.

  6. I guess after all these retraction watch exposures, I cannot accept certain things immediately. I was thinking about the time-line of this correction/errata. Abnormal science first published about this issue on December 05 and probably Dr.Zwriner might have informed the journal and the PI a week before or after the blog. Then the errata is already published in Feb 3 issue. During this time some of the experiments were repeated. I applaud this lab’s agressive nature of research. Remarkable!!

    1. with the lab size is it going to be that difficult to generate a blot?? The big labs / names have their ‘dirty little secrets’ which will never come out ..

  7. Reply to Aktfamily February 2, 2012 at 11:00 pm.

    I understand what you mean by “stupids”, but in English adjective needs something else to follow. Stupid ones, stupid people. Greek friends say “stupids”, which has the added bonus of sounding funny. The fact that “stupids” is intelligible does show the fexibility of English, and its ability to absorb and adapt. At one time adjectives did have endings, but these were lost. “Stupids” does add colour to the language. I make spelling msitakes all the time so it was not meant as a criticism, but as a welcome.

  8. Essentially this is a “do over” in which questionable data are being replaced with data from an apparently kosher “new experiment”. In general, my observation is that this isn’t an option thats available to most authors when these kinds of issues are uncovered. I can’t imagine this isn’t going to be an issue next time something like this happens at Cell (or in cases where papers have been retracted without the option of replacing questionable data).

  9. Seems like it’s time to start devising metrics for institutions involving scientific misconduct. Regardless of whether overt fraud was committed, I would still place this type of negligence well within the bounds of the misconduct category. Back to the metric issue – perhaps those institutions with particularly high rates of misconduct should be mandated by funding agencies to take corrective actions, in the form of additional training in research integrity. It certainly seems like Harvard and her affiliates could do more, given their boundless resources, to teach its trainees that above all else, sound science takes precedence over individual ambition. One also wonders whether this era of the “big paper,” consisting of massive datasets culled, probed, and validated by various camps across different institutions, intrinsically fosters this type of bad practice.

  10. http://www.cell.com/abstract/S0092-8674%2808%2901609-7

    This article has been retracted at the request of the Authors.Our paper reported the identification of a nuclear protein complex comprising DNA-PKcs, PIDD, and caspase-2 and characterization of its role in G2/M checkpoint maintenance, thereby providing insight into the functional significance of nuclear caspase-2. We recently identified errors affecting several figure panels where original data were processed inappropriately such that the figure panels do not accurately report the original data.

    Why couldn’t these authors just swap in some different data to replace the “errors”?

  11. @scotus: simple, they are not famous and not from big institutions..It is intersting that if you click on the above link provided, it takes you to the abstract of DNA-PKcs paper. It also gives a list of related papers which are retrated on the side bar….I found some interesting retractions there.

    1. @ressci They may be famous, but it is also possibly the case this is not a case of misconduct. If not then the correction is the right way to deal with it. It just depends on the facts and none of us know what the data are. I for one would not jump to the conclusion there is some giant conspiracy out there. There usually isn’t.

    2. @ Ressci
      Stowers is a big institute and David Chen is a famous name in DNA-PK and dna damage response field.
      As Jane said lets not jump to conclusions … may be there is more to it than the authors are disclosing in the retraction notice. If so, I hope they are doing something more about it internally

      1. thanks expostdoc, yes, I know whom you are talking about and have known them personally. but they are not as big as the person we are discussing here…..some of the investigations do not lead anywhere…I don’t know whether you recall the other herbal case…http://www.retractionwatch.com/2011/10/27/cancer-journal-retracts-herbal-medicine-paper-citing-misconduct-probe/ no news about this case and this person is enjoying his science even after this. Nothing will happen to any one – we are just putting our thoughts on to this blog.

  12. @Jane’s Addiction:Yes, I agree that it may not be a serious misconduct. We never know what has happened. But it involves two different publications though…we are all so pressed for publishing novel data – we might be under hypoglycemia and make mistakes…Agreed…Apologies would do.

  13. Let me ask each and everyone of you, in your professional carers, have any of you NEVER made a mistake of any sort ?

    If you have NEVER made any mistake in your life, then you may continue to criticize onthis blog.

    However, as you are human, ALL of you would have made some mistakes unintentionally.

    Please stop sensationilizing and dramatizing mistakes that other people make, simply because these mistakes are open to public scrutiny, and yours was probably not.

    1. anyone who has ever put a paper together as a FIRST author KNOWS that for these to be just mistakes is, in my opinion, inconceivable. I know all my images and figures by heart and to publish the same one twice boggles the mind. Now add into the mix the myriad other authors on these papers and things get cloudy. the buck stops at the corresponding author. when clinical trials and patents are the basis of primary research, especially from Hahhhvard, there is no excuse

  14. Dear scienceobserver February 5, 2012 at 9:20 pm,

    I go back to the first comment which appears under the section


    Gabor Hrasko

    October 17, 2010 at 4:22 am

    “Science and pseudoscience differs not in that one contains fraud and bad practices and the other does not, but in that in science there is at least a willingness to clear up the mess sometimes. Retraction Watch is a brilliant idea to present this effort for self correction and openness of science”.

    Gabor Hrasko
    President – Hungarian Skeptic Society

    This is what we should always be doing: “science there is at least a willingness to clear up the mess”.
    Concentrate on this function. When people ignore this function others do become upset.

    1. good response Bernard…i think this kind of response was frequently pop up when the issue is personally relevant. I was following earlier abnormal science and i noticed that person got affected (or his/her associates, this was my suspicion) was responding in a rather threatening manner. Scienceoberver was really disappointed as there was some strong response on this blog: http://www.retractionwatch.com/2012/01/31/md-anderson-investigating-researcher-bharat-aggarwal-over-images/

  15. @ Ressci integrity: Scientists publishing the data with their name, even though that is manipulated or with errors….so they are open…but we here publishing or posting criticism (sometimes blind criticism) hiding our name. …..just think who is the real fraud?

    1. Excuse me? Scientists publishing manipulated data with their name on it? You are apparently defending scientists who knowingly commit misconduct – because they put their names on their papers? Of course they put their names on their fraudulent papers – that’s the point isn’t it? They just hope they aren’t caught.
      “Scientist” should do some more reading here at RW – about the numerous scientists who author papers filled with fabricated data, expecting to get credit and NOT get caught. (See Mori, Potti, etc.). These people are certainly not “open” when the investigation begins… Noted that “scientists” did not disclose his/her name.

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.