Figure duplication kills cell death paper

cdd_cimageA pair of researchers at the University of Maryland have retracted a paper in Cell Death & Differentiation after it became clear that one of the figures had been duplicated from an earlier paper.

Here’s the notice, dated December 13, 2013, for “INrf2 (Keap1) targets Bcl-2 degradation and controls cellular apoptosis,” by Suryakant Niture and Anil Jaiswal:

Although majority of the data are reproducible and conclusions do not change, as they are supported by several experiments, the corresponding author has decided to retract this article in its entirety due to reuse of Figure 5a from other publication by the first author.

We’re not sure what prompted the retraction (and have asked the corresponding author), but we do know that the pseudonymous Clare Francis contacted the journal on July 5 of last year with a number of concerns, including some about Figure 5a. Similar comments have been left on PubPeer.

The paper has been cited 18 times, according to Thomson Scientific. The most recent citation was by the authors of the now-retracted paper, in a review published a few weeks after the retraction. While the review was no doubt in press before the retraction occurred, it’s hard to imagine it was done and dusted before the decision to retract.

The work was supported by the NIH.

16 thoughts on “Figure duplication kills cell death paper”

  1. I think this is the absolutely correct thing to do. CD&D has set the bar really high: one duplicated figure = retraction. This is significant. The publisher, Nature Publishing Group (NPG), is NOT a member of COPE:

    That, to me, is a fact of even greater significance because it suggests two possibilities:
    a) NPG does not need this “ethics” group to validate its publishing ethics.
    b) COPE does not hold its “members” (aka clients) to the high levels of ethics that NPG expects of its journals and journal editors.
    The important question to ask here is, when will COPE come out and define some clear ethical bars for rejection, for example, is one duplicated figure the bar for rejection? By association, is one duplicated figure the equivalent of one paragraph of duplicated text, or one duplicated table?

    In fact, I think that every single publisher has to now start to set out the exact parameters that would make a paper eligible for rejection. Enough of these back-room deals by “ethicists” in secret panels and lawyers that are imposing pseudo-ethical rules on the scientific community. Although I am a firm critique of scientific fraud, the current lax, undefined and variable rules by each editor, journal and publisher is now starting to get me really irritated. Why don’t we start at the top? Can an executive or PR manager from Elsevier Ltd., Springer Science + Business Media, Taylor and Francis, and Wiley-Blackwell simply not come to Retraction Watch and state to the scientific community: “The bar has been set as follows: one duplicated figure, table, paragraph of text that measures >2% of the entire text, or data point is reason for retraction, independent of the excuse offered by the authors”. One rule for an across-the-board application to all their journals? Why is this so complicated? Does it really take so many editors, “ethicists” and lawyers to come out with a simple rule that we, the scientists, can then come to respect and abide by? We hold them accountable, too. After all, it is evident that their PR teams are monitoring this and a handful of other similar blogs, so why fear being decisive? It is precisely the messy handling by publishers, especially this top echelon, that is now also causing anguish among scientists. It is precisely because of their messy business that claims ethics but cannot even quantify it, that is causing scientists to move in droves to other publishers, and to create a tsunami of open access journals and publishers, many of which are academically predatory and fraudulent. In the last 1-2 years, these two four publishers have now started to try and “clean up shop”, with this wave of retractions, but this is not the correct way to go. You have to quickly establish a correct set of rules, you have to have a team of professionals that makes these decisions publically, openly and respecting established rules, much like scientists would expect of a peer review by a team of editors on an editor board. You must have one senior ethicist cum lawyer who then takes public responsibility for that decision and is held accountable, much like an editor-in-chief must be held accountable for the papers he/she approves under his/her reign at the top. If these top four publishers can assure fixed and quantifiable ethical guidelines that they decide, and implement, and not through some fuzzy “ethics” group like COPE, then I am quite confident that scientists will come back in droves, because trust will have been re-established. Science is in crisis, as is science publishing, partly because scientists have been held hostage in the publishing process by these groups. It is time that we apply, as scientists, more pressure on the publishers, because finally it is they who hold the POWER of retraction, not the scientist that dabbles in academic or publishing misconduct.

    1. Much was written about how high the bar is set by Nature Publishing Group and COPE (Committee on Publication Ethics). There is an article in J. Cell Biology 2004; 166:11–15 about image manipulation. So COPE does have some guidelines. It is left to the affiliate journals to set how high their bar should be.

      May be this example is worth examining. In the J Cell Sci 2009, 122: 4452, Fig 9B, the authors had demonstrated cleavage of Caspase 3 in MCF-7 cells, when MCF-7 cells do not express Caspase 3 (Breast Can Res Treatment 2009, 117(1): 219). When this issue was exposed, the authors published an erratum stating that the Fig 9B was from Hepa-1 cells and not from MCF-7 cells, indirectly explaining the DNA fragmentation noticed in MCF-7 cells was due to Caspase 3 activity in Hepa-1 cells, which surprisingly was acceptable to the journal. On comparison of PKCd amounts in MCF-7 cells (PKCd+/+) to that of Hepa-1 cells in Fig 5A (notice very little amount; also refer Fig 3A), this suggests that the PKCd WB in Fig 9B is indeed from MCF-7 cells as originally reported and not from Hepa-1 cells as against what is stated in the erratum. I believe that the authors replaced MCF-7 cells with Hepa-1 cells to create the Caspase 3 blot (without mentioning the switch) to demonstrate a result they wanted, and sent an erratum statement which again is not true because the bottom three blots of Fig 9B are still from MCF-7 cells. also has some entries on this paper and others by the same lab: . Of particular interest is the apparent similarity of Fig 4 Bax panel of Cancer Res 2008 Oct 1, 68(19): 7915-22 to the p53 panel (first 4 lanes) of Fig 4B of Clin Cancer Res 2009 March 1, 15(5): 1534-42.

      Another area of attention is how graphs are created and statistical analysis were done. An example, Cancer Res 2006, 66(17): 8421-29, Fig 3D. Pay attention to the creation of the graph and the distribution of upper and lower halves of SD bars, also to their floating or shifting nature (Fig 5C), zoom in. Now use your analytical skill and scan over other publications starting with rest of the figures of the same 2006 Cancer Research.

  2. Entry at Pubpeer on a publication by the editor-in-chief of Cell Death and Differentiation.
    How to explain it? There must be an explanation.
    Editor-in-chief Cell Death and differentiation.
    Pubpeer entry.
    Imgur image as part of Pubpeer entry.
    Original article.
    Blood. 2009 Jun 18;113(25):6498-9; author reply 6499-500. doi: 10.1182/blood-2009-02-203174.
    p73, miR106b, miR34a, and Itch in chronic lymphocytic leukemia.
    Rivetti di Val Cervo P, Tucci P, Majid A, Lena AM, Agostini M, Bernardini S, Candi E, Cohen G, Nicotera P, Dyer MJ, Melino G.
    Correspondence: Prof Gerry Melino or Martin J. S. Dyer, MRC-Toxicology Unit, Hodgkin Bldg, PO Box 138, University of Leicester, Lancaster Rd, Leicester, LE1 9HN United Kingdom.
    PMID: 19541840

    1. Yes, this is a good observation. However, in this case, no action has been taken, the PubPeer comment sits idle and silent. Why don’t you simply contact Dr. Melino and point to the RW stiry and the PubPeer entry and request a comment and explanation for the (apparently) identical banding? You can use an anonymous identity. And if Dr. Melino doesn’t respond within a few days, well, then forward that request to the entire editor board. One would expect at least one member of the board to have a conscience and react. Related to this, a hypothetical question. If an editor on an editor board of a journal that carries an impact factor has a clear act of duplication, even if the paper is not retracted, shouldn’t this be sufficient reason to call for the resgnation of that individual from the editor board? Wouldn’t alowing an editor that has a clear case of duplication (data, text, figure, or table) be sufficient ethical reason to call for a resignation? I don’t see how two sets of values can be compatible and imposed on the authors, but not on the editors who are supposedly the quality and ethics gate-keepers of that journal.

  3. Hsp90 interaction with INrf2(Keap1) mediates stress-induced Nrf2 activation.
    Suryakant K. Niture and Anil K. Jaiswal
    VOLUME 285 (2010) PAGES 36865–36875

    This article has been withdrawn by the authors.

    © 2014 by The American Society for Biochemistry and Molecular Biology, Inc.

  4. It is not just figure duplication that killed Cell Death Paper. It is the author’s Retraction Statement “reuse of Fig 5a from other publication”, “conclusions do not change” that misled to all these confusion and for comments like “one duplicated figure = retraction”, “Is one duplicated figure the bar for rejection?”. There is more than reuse of Fig 5a, which the authors do not want to convey. After reading the observations given below, it is left to the journal to edit the retraction statement, if convinced.

    To understand the facts given below a copy of the following publications in hand would be helpful. 1) CDD 18: 439-51, 2011 + its Supplementary figures; 2) JBC 284(20): 13856-68, 2009; 3) JBC 285(47): 36865-75, 2010; 4) JBC 286(52):44542-56, 2011 and a knowledge in interpreting Immunoprecipitation (IP) expt. results.

    Fig 5a (IP figures) of CDD 18. Compare this figure to JBC 284(20) Fig 3A. The only difference is added Bcl2 blots in CDD. So the statement is politically correct, reuse of Fig 5a from other publication. But are the results of JBC 284 Fig 3A themselves scientifically correct? Why lane 2 of bottom most panel does not contain IgG HC or LC although it is labeled as Con IgG (IP)? By excluding Con IgG IP in IP experiments (but representing having been done), Con IgG IP lanes of all the panels of that figure (and in many of their publications) looked pretty clean and could avoid showing non-specific interactions. This pattern of omission of Con IgG IP in IP expts, but represented as having been done could clearly be seen in JBC 286(52), Fig 5A Right, top WB:Bcl-xL panel, lane 2, IP IgG (IgG LC absent in lane 2, the lower thicker band). There is even an evidence of erasing these type of “non-specific” band in Con IgG IP lane (same JBC 286(52): Fig 5C, IP panels, 2nd from bottom, IP:PGAM5L-WB:Bcl-xL, lane 2, above the band labeled IgG, zoom in).
    The creation of Bcl-2 western in CDD 18 Fig 5a IP panels. For this refer to JBC 284(20) Fig 5B. Top 2 panels (IP:INrf2-WB:PTMα; and IP:INrf2-WB:INrf2) and second panel from bottom (IP:PTMα-WB:PTMα). Cut out the first 4 lanes, do little bit of stretching/shrinking and label them as Bcl-2 western blots. This way the authors ‘demonstrate’ IP/co-IP of Bcl-2 with INrf2, Cul3 and Rbx1 antibodies to claim later targeting of Bcl-2 by INrf2 for ubiquitination/degradation. The implication we will discuss below under “conclusions do not change”. Obviously this is also a ‘reuse of figure’, but how was it reused? This is where scientific “ethics” comes in reporting facts.
    How ‘reuse’ of CDD Fig 1b Bcl-2 blot in Fig 1e after flipping vertically in a way they wanted was reported in Pubpeer with links to and also reuse of β-actin blot in Fig 4B from Fig 1B.
    Genesis of CDD Fig 3e, Input-WB GFP. Compare its similarity to JBC 285(47): Fig 2B, IP:GFP-WB:GFP. Remember CDD Fig is an Input and JBC Fig is an IP. (These two are different gels created from same set of lysates. In IP one expects more protein to be pulled down). Conclusion: IP expts. were created just by loading cell lysates without doing IP. Now to trace the common source of the lysates for these expts. For this you need to refer to JBC 284(20) Fig 1B. It is a rabbit reticulocyte TNT expt. lysate. Compare this figure to JBC 285(47): Fig 2B Input. TNT reticulocyte lysate has now become Hepa-1 cell lysate. Compare the amount of each band to the nearby IP:GFP-WB:GFP blot. They are in equal proportions. The IP blot was created by just loading the same TNT lysates and reported as having been done in Hepa-1 cells without doing an IP.
    Now examine Fig 3e of CDD, IP:GFP-WB:GFP (2nd panel on the right). Stretch it vertically and compare it to JBC 285(47): Fig 2B Right, IP:GFP-WB:GFP. The figures are identical. Of course this is also a reuse of figure from other publication. However, compare how the Con IgG IP lanes were labeled in the two blots (in one it is on the left and in the other it is on the right. Both are empty spaces of the film!). Also notice Input lane in JBC 285 (47): Fig 2B IP:GFP-WB:GFP. No signal! This is also an empty space on the film. (This JBC manuscript is now withdrawn; also contained numerous other fabricated data).
    “Conclusions do not change”. Go to Supplementary Fig S2a WB:V5 of CDD. Darken it and place it next to Fig S2c bottom panel, IP:V5-WB:V5. Both are same. Again an IP blot was presented without doing IP. So what about lane 1 in Fig S2c top panel? It is an empty space. What about IPs claimed to have been done in the rest of lanes? No presence of IgG HC or LC in the whole blot although it covered from 150 kDa to 25 kDa. So the ubiquitination pattern (WB: Ub) shown as specific to V5 tagged pulled down proteins is not true as no IP was done. Similarly no IP was done in Fig S2d. No presence of IgG HC or LC in the whole blot. Moreover, in lane 3 ubiquitination is observed. The difference between lanes 2 and 3 is addition of empty pcDNA vector in lane 3 in this in vitro assay.
    The same observation holds good to results shown in main text Figs 2a, b, c (IP panels) where there are no indication of an IP done. Special note: The refined current technology to eliminate detection of IgG HC or LC in IP expts. was not available at the time these experiments were conducted nor was it mentioned in Materials and Methods. Under these observations there is no convincing data to prove that INrf2 targets Bcl-2 degradation by ubiquitination and the conclusion is questionable.

    1. In reply to Robert April 27, 2014 at 1:54 pm

      Robert many thanks for your thorough analysis of the 4 papers. Your final sentence is the oppositie from “conclusions do not change”.

      1. Sorry for the confusion. The author’s statement “conclusion do not change” is questionable.

        The now retracted 2nd manuscript (JBC 2010, 285(47): 36865) is another good teaching material for investigative journalism. Now that clues how to analyze WB and IP data were given, this paper would be everybody’s take home assignment. We already had analyzed Fig 2. More are on Pubpeer on Figs 5, 7 & 8. You need to zoom in on the bands. There are more on Figs 3 & 10 and the one on Fig 6B is for extra credit (a knowledge to interpret IP result is essential).

        1. JBC 2010, 285(47):36865, Fig 3. It is amazing how perfectly bands are created in IP:Flag-WB:V5 blot, band deltaCTD from IP:V5-WB:V5 and band deltaCTD-deltaMD from IP:V5-WB:Flag (copied and pasted). Only problem their relative migrations are not correct. Zoom in. This is scary. What is real now?

  5. Why the retraction watch or journalist is not asking Dean,School of Medicine, University of Maryland, Professor Anil Jaiswal (410-706-2285) and Dr. Suryakanta Niture (410-706-2284) for clarification? Also to the NIH Funding agency is providing the following resources for this lab to conduct research per year from last 20-25 years. whats the out put

    5 R01 ES007943 18
    NIEHS $334,125

    5 R01 ES012265 11
    NIEHS $330,785

    5 R01 ES021483 02
    NIEHS $341,921

    5 R01 GM047466 22
    NIGMS $283,946

    1. The University of Maryland at Baltimore [UMB]–which pompously insists on being called “The Founding Campus”– appears to be standing behind both faculty members.

      All three retracted papers are still on the official resumes of Dr. Jaiswal and his research associate Dr. Niture in their UMB Faculty Profiles [as of 8/15/2014]–even though it is UMB policy to delete retracted papers from faculty resumes–and Dr. Jaiswal is still co-director of Graduate School’s Program in Toxicology.

      I am currently a doctoral candidate in this program and am stunned that the students have not been informed, even though UMB must have known of the issues in Jaiswal’s papers since at least the first retraction in Dec 2013.

      Dr. Jaiswal’s three retractions and growing list of other suspect papers are apparently not considered significant enough to take away any of his faculty responsibilities. They have not cost him his UMB job or any of his NIH grants, and they don’t appear to have cost UMB anything since NIH has not requested any refunds.

      Dr Jaiswal published another 5 papers with Dr. Niture since 2009.

      How long will the the Foundering Campus take to appoint someone else to be co-director of the Program in Toxicology? I’ll post an update if it happens before I graduate!

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.