UCL finds errors in work by biologist Cossu, but no “deliberate intention to mislead”


A cell biologist at University College London (UCL) who has had one paper retracted and another corrected has been cleared of misconduct by the university.

The news, first reported by Times Higher Education, comes after a retraction of a paper by Giulio Cossu prompted by pseudonymous whistleblower Clare Francis that we wrote about in January.

Here’s the full text of UCL’s statement on the investigation:

The panel set up by UCL’s Research Governance Committee to look into the queries raised by anonymous correspondent ‘Clare Francis’ into research conducted under the supervision of Professor Giulio Cossu has completed its inquiries. Because the queries were made anonymously, it was considered impracticable to pursue them through UCL’s usual procedure for investigating allegations of misconduct in academic research. Nevertheless, UCL’s commitment to good practice in research obliged us to look into the queries raised.

The allegations made by ‘Clare Francis’ concerned eight research publications. The panel found no substance to the large majority of these allegations. With respect to the few remaining allegations, the panel found that the conduct and presentation of the research in question did not always meet the standard of good practice expected of UCL researchers, and enshrined in UCL’s ‘Code of Conduct of Research’. Where specific errors were detected, the panel recommended that, where possible, corrections should be made to the published papers. Overall, the panel concluded that there was no deliberate intention to mislead on the part of Professor Cossu and that no activity had taken place that could be considered as research misconduct.

Accepting the panel’s findings, Professor Cossu has acknowledged the spirit and details of the report. He accepts that there are some errors present in previously published papers, a fact that he sincerely regrets. He will work actively with colleagues at UCL, and implement processes to ensure to the best of his ability that such errors do not occur in the future.  As regrettable as they are, the investigation found no indication that the errors altered the fundamental conclusions of these studies, many of which have been confirmed by independent research groups.

We’re not sure why it is “impracticable to pursue” anonymous complaints, as we’ve argued before. We’ve asked UCL why the full report is not being made available, and which papers will be corrected. We’re also curious why Cossu has had a paper retracted if “the investigation found no indication that the errors altered the fundamental conclusions of these studies.” We’ll update with anything we learn.

The investigation is the third we’ve covered involving UCL. The first was into the work of Jatinder Ahluwalia, and the second into that of Assegid Garedew.

Please see an update on this post.

103 thoughts on “UCL finds errors in work by biologist Cossu, but no “deliberate intention to mislead””

  1. As the blog post points out this fact of the matter which appeared during the investigation


    does make “the investigation found no indication that the errors altered the fundamental conclusions of these studies.” seem counterfactual.

    Another point is that I do not believe that anybody has confirmed G Cossu’s work on treating muscular dystrophy in dogs for which a correction has been issued.


  2. I think the comment on the JCB paper is an unwarranted snipe at UCL, since that paper had nothing to do with UCL. Cossu was in Milan at that time, so why should UCL investigate that paper?

  3. Regardless of where the JCB work was done, a condition of employment at a UK university and holding RCUK grants is ethical conduct. Five of the 6 figures in the retracted JCB paper that were found to have problems concerned the re-use of data to describe a different experimental condition. This cannot be the result of an “error” and someone is captain on the ship and has to take responsibility. Perhaps we need an addition to the “Thesaurus of euphemisms” I proposed in a recent blog post.

    1. Fair enough, but UCL has no reason to investigate a paper that is not linked to its institute. Besides that, he wasn’t the primary or senior author on that paper.

      1. I haven’t looked at any of the papers in questions, but he seems to have moved to UCL reasonably recently less (ie 2 years ago or less).
        Taken to the logical conclusion all you need to do when you run into issues is move institution and then the slate is wiped completely clean.

        1. Well, that’s not what I said. It’s the university of Milan that would have to investigate that paper, and if fraudulent handling is found, the authors should be punished (or maybe already have, who knows?).

  4. The report doesn’t say that “it is “impractable to pursue” anonymous complaints”. It states that “it was considered impracticable to pursue them through UCL’s usual procedure for investigating allegations of misconduct in academic research” (my italics). Clearly UCL doesn’t consider that it is “impractical to pursue anonymous complaints” at all since they clearly did pursue this anonymous complaint.

      1. I expect UCL included that sentence because they wished to convey the nature of the complaint and why it was investigated in the manner that it was…

  5. Doesn’t anybody read C.P.Snow any more. One has to avoid unintentional errors as well as intentional ones.
    Admitting that this is impossible, one should certainly try- wide consultation with ones colleagues (including use of skype and telephones!), longer periods for reviewers to think about the papers they are reading, and perhaps blogs devoted to possible sources of errors in different kinds of work might help. How about blogs to inform reviewers of what to look for when reviewing specific techniques?

    I suggest that scientists should set higher standards for the papers they cite. I cannot imagine citing anything from an author who has already retracted 50 papers! but we might also refuse to cite work in journals which dont give clear reasons for retractions.

    And while I am suggesting things, how about paying reviewers? $250 for 2 days spent on a paper would certainly help an advanced grad student.

    1. In my experience, giving me more time to review a paper just means I start reviewing it later, and I see no reason others do not do the same. Getting paid is great, too, but how do you assure someone really used that much time on the paper?

      To continue being a bore: I don’t see any reason to punish authors for something the journal does or does not do. I will continue to cite papers from JBC if the work looks solid (or not, with the appropriate remark about that), regardless of their, now-solved it seems, vague retraction notices.

      A repository of typical problems with techniques/methods/approaches would be useful, though!

    2. Please correct me if I am wrong, but I am fairly sure that a failure to cite relevant papers is considered a form of scientific misconduct.

      1. I don’t know that this is a reply to me- but it IS an increasingly difficult question. It used to be that you tried to quote everything relevant, or a review, or the last paper in a series. You certainly didn’t choose according to who you admired though you might give them more weight. However now there are so many second rate journals that you have to choose. A real problem arises if you know a paper is reporting data that cannot be true and yet hasn’t been retracted. This actually arises in my area of research – and I have a choice of saying it has now been proven to be wrong, or I ignore it. In fact I have done both. Elaine Newman

      2. PWK. I coined the term “snub publishing” to describe this phenomenon (http://www.globalsciencebooks.info/JournalsSup/images/2013/AAJPSB_7(SI1)/AAJPSB_7(SI1)35-37.pdf). I think the problem is very serious, at least in the plant sciences. One of the reasons, if I may be so frank, is the total fault of editors and publishers who literally force the authors to cut back on what they might perceive as being redundant references, or indicate that there is a limit to the number of references in a reference list. Of course, there are those authors who purposefully exclude authors or overly include a set of others, for political or other reasons. Hope my definition and scale allow this problem to be quantified further.

  6. OK – some contrition from the author and a very vague acknowledgement that there was ‘sloppy science’ – but broadly this looks like a pretty typical British institutional fudge. The number and type of issues afflicting the JCB paper clearly indicate misconduct by someone in the group even if it was not directly Cossu himself; to deny this suggests a superficial examination of the issues.

    The statement that there was ‘…no indication that the errors altered the fundamental conclusions of these studies, many of which have been confirmed by independent research groups.’ appears inappropriate for three reasons. Firstly as Fernando says, this doesn’t square with the JCB retraction – that means invalid results. Secondly, I don’t buy the relevance of work being confirmed by other groups – bad work often gets ‘confirmed’ by other bad work. Life is simply not as simple as this panel makes out.

    Thirdly and most fundamentally, the investigation panel are not the right people to be making statements about whether a finding has been confirmed by others in the field or not – they should be independent experts with a good understanding of the broad scientific method, whose job is to look for misconduct in a critical and objective way. The UK Research Integrity Office (UKRIO) has no powers, but it has at least produced a clear policy on how to investigate misconduct. This says nothing about the panel opining on whether findings have been ‘confirmed’ by other groups outside the university. Their job is simply to tell us if there was misconduct and to recommend correction / retraction of certain articles.

    Put another way: ‘The work was unreliable, but its conclusions seem OK so all’s well that ends well…’ It’s amazing how often this concept is used to excuse invalid research.

    They did better than some, but in my view this committee would score ‘4/10, could do better’.

    1. In agreement with amw May 9, 2013 at 12:32 pm and to restate:
      I do not think that anybody has confirmed the work on treating muscular dystrophy in dogs.

      Nature. 2006 Nov 30;444(7119):574-9. Epub 2006 Nov 15.
      Mesoangioblast stem cells ameliorate muscle function in dystrophic dogs.
      Sampaolesi M, Blot S, D’Antona G, Granger N, Tonlorenzi R, Innocenzi A, Mognol P, Thibaud JL, Galvez BG, Barthélémy I, Perani L, Mantero S, Guttinger M, Pansarasa O, Rinaldi C, Cusella De Angelis MG, Torrente Y, Bordignon C, Bottinelli R, Cossu G.
      San Raffaele Scientific Institute, Università Vita e Salute, Stem Cell Research Institute, Via Olgettina 58, 20132 Milan, Italy.
      PMID: 17108972
      Erratum in
      Nature. 2013 Feb 28;494(7438):506.

      An earlier claim to have corrected dystrophin deficiency in mice by one of the authors of the Nature paper
      looks decidedly shaky. Hint: look at figure 4 lower two panels and the upper panel.


      Hum Mol Genet. 2000 Jul 22;9(12):1843-52.
      Transplacental injection of somite-derived cells in mdx mouse embryos for the correction of dystrophin deficiency.
      Torrente Y, D’Angelo MG, Li Z, Del Bo R, Corti S, Mericskay M, DeLiso A, Fassati A, Paulin D, Comi GP, Scarlato G, Bresolin N.
      IRCCS Ospedale Maggiore Policlinico, Milan, Italy.
      PMID: 10915773

      1. How come some people do not like it when things which are incredibly unlikely to happen by chance are pointed out? Are they aristocratic against mathematics and physics? The pattern of the bands and associated dots, flecks, etc., in some of the lanes in figure 4 Hum Mol Genet. 2000 Jul 22;9(12):1843-52 are identical. The different lanes represent different samples/reactions. Even if it were the same reaction running some in one lane and some in another lane should not give you identical bands.

      2. I agree these bands have been copied and pasted. People have previously pointed out that flecks on the camera lens or other piece of equipment can cause such issues but these bands are also identical in their DNA pattern. It’s quite a skilled job – in the bottom panel they seem to mixing and matching upper and lower bands so it is very hard to spot a pattern. And there is very little evidence of splice artefacts.

        These people (to be clear, this is not Cossu but a co-author) were clearly fabricating their gel images. Thirteen years on, of course no one will do anything about it, but it makes the point that UCL are wrong to refer to other work which they haven’t examined as evidence to support Cossu’s conclusions.

        Fraudulent work attracts fraudulent confirmations like rotten wood spreads. In each field there is usually a core of critical scientists who know what is right and what is not, but for non-experts and students this kind of aggregated fictional science is impossible to analyse – a nightmare.

    2. I repeat: the JCB paper was not from UCL. UCL cannot investigate potential misconduct in a paper that is not connected to UCL. Possibly the University of Milan has done so.

      1. We are talking science.
        The issue is now the credibility of Hum Mol Genet. 2000 Jul 22;9(12):1843-52

  7. Stem Cells. 2009 Jan;27(1):157-64. doi: 10.1634/stemcells.2008-0503.
    Skeletal muscle differentiation of embryonic mesoangioblasts requires pax3 activity.
    Messina G, Sirabella D, Monteverde S, Galvez BG, Tonlorenzi R, Schnapp E, De Angelis L, Brunelli S, Relaix F, Buckingham M, Cossu G.
    Stem Cell Research Institute, H. San Raffaele Scientific Institute, Milan, Italy.
    PMID: 18845762

    Figure 1A.
    Pay attention to D16, D15 and D63 lanes in the Pax3 panel and the bands in the D63, D15 and D16 lanes of the beta-tubulin panel.

    Figure 1B. Where is the background?

    Figure 1C. Which of these things belong together? How many birdies, how many frogs?

    1. Fernando, I love your posts, but you really should just spell it out sometimes. Let me start, lanes 1, 2 and 3 of the Pax-3 and b-tubulin blots are obviously mirror images of one another. This particular blot has several…err…”defining characteristics”.

      1. In reply to Freeloader May 9, 2013 at 3:13 pm

        Thnaks. I’ll bear that in mind. I thought that people might like to join up some of the dots themselves.

      2. In reply to Freeloader May 9, 2013 at 3:13 pm

        What does that mean when you see mirror images, or things turned up-side down? Is it something of a pagan hangover from the god Janus.

        You can see it in this production: Blood. 2000 Mar 15;95(6):2084-92. PMID: 10706878

        Compare figure 1A and top panel figure 5.

        By the way, top physicists in Berlin are trying to explain it (away). “Very interesting results”.

        1. The panel stated that the ‘…conduct and presentation of the research in question did not always meet the standard of good practice expected of UCL researchers, and enshrined in UCL’s ‘Code of Conduct of Research’.

          Without a report, it is reasonable to ask (given the large number of issues pointed out by Fernando and others), what this means – did the panel see exactly the sort of issues that Fernando points out? If they did, they must have chosen to define them same issues as sloppy science.

          Former US Secretary of Defense Donald Runsfeld tried to suggest that forcing prisoners to stand for four hours was not torture since he claimed to stand for 8-10 hours per day. If you don’t like what someone did, you can simply find a way of redefining it based on another set of rules.

          Repeated mirroring and copy / paste across a wide range of papers is strong evidence of fabrication. Once you start creating your gels like this, and you get publications in Nature and other high-profile journals, you don’t stop (the formula works); we’ve seen countless cases like this.

          Practically, there should be at least some answers when we see what the corrections actually turn out to be. I suspect RW will monitor.

  8. I am not sure what these investigation say about the institutions and the Journals. Fool me once, shame on you, fool me twice, shame one me, fool me, oh wait, I stop counting, shame on all of us.
    I do not get, how can we consider duplicating figure and erasing bands an honest mistake. That does not even pass the basic smell test

    1. In reply to Schmuck May 9, 2013 at 9:40 pm

      You mention the smell test. Here is the sight test.

      First paper that looked odd.

      PLoS One. 2008 Sep 16;3(9):e3223. doi: 10.1371/journal.pone.0003223.
      Magic-factor 1, a partial agonist of Met, induces muscle hypertrophy by protecting myogenic progenitors from apoptosis.
      Cassano M, Biressi S, Finan A, Benedetti L, Omes C, Boratto R, Martin F, Allegretti M, Broccoli V, Cusella De Angelis G, Comoglio PM, Basilico C, Torrente Y, Michieli P, Cossu G, Sampaolesi M.
      Translational Cardiomyology, Stem Cell Institute Leuven, KULeuven, Leuven, Belgium.
      PMID: 18795097

      Figure 2.
      Figure 2E.
      Myostatin panel. Bands in sk Mus and E10.5 lanes are distinctive but resemble each other.
      Follistatin panel. Vertical change in background between the -RT and PM lanes. Suspect splicing.
      The bands in the DM, DC and E10.5 lanes are not the same, but have a similar wavy appearance.
      IGF-I panel. Bands low resolution, made of about 6-7 horizontal light strips. Do not look natural.
      GAPDH panel. Bands resemble interlocking railway trucks.

      Figure 3 .
      MagicF-1 panel. Vertical change in background between the 3 w and H20 lanes even though this separates the bands 1 w, 2w and 3 w from the Marker lane on the left.
      HGF pane. Dark vertical streaks between the 3 w and H20 lane with same result as above.

      Figure 4.
      Figure 4C. MAGIC panel. Low resolution. Some of the bands which are distinctive, but not the same, resemble each other.
      GAPDH panel. Low resolution. Some areas of lower resolution. Some of the bands which are distinctive, but not the same, resemble each other, including some of the sample bands resembling the marker bands.

      Figure 5.

      Figure 5A. WT panels. The architecture of the upper 3 d panel looks very like the lower 7 d panel.

      Figure 6.

      Magic-F1 panel. Low resolution. Bands which are not the same, but not that much different.

      Suggested looking at more papers.

      1. J Clin Invest. 2004 Jul;114(2):182-95.
        Human circulating AC133(+) stem cells restore dystrophin expression and ameliorate function in dystrophic skeletal muscle.
        Torrente Y, Belicchi M, Sampaolesi M, Pisati F, Meregalli M, D’Antona G, Tonlorenzi R, Porretti L, Gavina M, Mamchaoui K, Pellegrino MA, Furling D, Mouly V,Butler-Browne GS, Bottinelli R, Cossu G, Bresolin N.
        Stem Cell Laboratory, Department of Neurological Science, Instituto di Ricovero e Cura a Carattere Scientifico Ospedale Maggiore Policlinico, Centro Dino Ferrari, University of Milan, Italy.
        PMID: 15254585

        Figure 3.

        Figure 3C. I suspect that many panels are composites.
        I suspect that the first lanes of the CD34, PAX7, M-Cad, MYOG panels have been spliced in.
        You can see vertical discontinuitites in background.
        In the MYF5, MYOD and beta-actin panels the background from the right seems to lie over part of the first lane.

        Figure 3F. The lanes of the upper two panels are very likey spliced together. Vertical changes in background.

        Figure 4.

        Figure 4E.
        I suspect that the bands in the M-Cad and MYF5 panels are the same. Many of the details in the backgrounds line up.

        Figure 7.

        Figure 7F. I suspect the bands in the beta-actin panels are repeats. They look like the bands in lanes 2,3 and 4 of the beta-actin panel of figure 4E, yet the shared panels above them look different.

  9. No deliberate intention to mislead? As pointed out by Fernando Pessoa in his February 28, 2013 comment (http://www.retractionwatch.com/2013/01/22/clare-francis-scores-a-bullseye-journal-of-cell-biology-paper-retracted-for-image-manipulation/) the Nature article contains FALSE data!

    Not only the VIRUS and VACCINE MyHC panels were “unintentionally duplicated” but the same panel has been used to create the VARUS MyHc lower panel.This can’t be unintentional because the panel has been intentionally twisted and the first lane if VARUS became the second lane of VIRUS-VACCINE MYhC panel (http://www.nature.com/nature/journal/vaop/ncurrent/full/nature11976.html). Are the Nature editors and UCL experts blind?

  10. J Cell Biol. 2006 Jul 17;174(2):231-43. Epub 2006 Jul 10.
    Complete repair of dystrophic skeletal muscle by mesoangioblasts with enhanced migration ability.
    Galvez BG, Sampaolesi M, Brunelli S, Covarello D, Gavina M, Rossi B, Constantin G, Torrente Y, Cossu G.

    Stem Cell Research Institute, San Raffaele Hospital, 20132 Milan, and Department of Experimental Medicine, Human Anatomy Institute, University of Pavia, Italy


    Figure 6.

    Right panel (all may not be well with the middle panel too). This is puzzling me.
    Many short, horizontal lines with dark dots.
    I think that there is aa repeat in in the background in the left (MOCK) lane.
    About 1/6 the way down the panel, near the left side there is a “v” with grey dots above it and a grey patch below it. A constellation of very similar marks appears about 1/3rd the way down the panel and a bit further to the right. I think that the repeated area may be more than this.

    1. Dev Biol. 2012 May 1;365(1):91-100. doi: 10.1016/j.ydbio.2012.02.015. Epub 2012 Feb 18.
      Noggin recruits mesoderm progenitors from the dorsal aorta to a skeletal myogenic fate.
      Ugarte G, Cappellari O, Perani L, Pistocchi A, Cossu G.
      Division of Regenerative Medicine, San Raffaele Scientific Institute, Milan.
      PMID: 22370001

      Figure 7.

      Figure 7A. Top (Noggin) panel. The bands in the C and in the SiRNA lanes are both straight and at angles to each other. They do not bend. How can they be running in the same gel?

      1. Sci Transl Med. 2011 Aug 17;3(96):96ra78. doi: 10.1126/scitranslmed.3002342.
        Stem cell-mediated transfer of a human artificial chromosome ameliorates muscular dystrophy.
        Tedesco FS, Hoshiya H, D’Antona G, Gerli MF, Messina G, Antonini S, Tonlorenzi R, Benedetti S, Berghella L, Torrente Y, Kazuki Y, Bottinelli R, Oshimura M,Cossu G.
        Division of Regenerative Medicine, Stem Cells and Gene Therapy, San Raffaele Scientific Institute, 20132 Milan, Italy.

        Figure 2B. The bands in the middle murine dystrophin panel and the right Human dystrophin panel look similar with the indentations on the left and right.

        Figure 3D. The bands in lanes 3 through 5 look very similar. Pattern of indentations very similar.

        Figures 4B and 4C. No background between the bands, which are tightly cropped.
        There is some background in lane 8 in the right DYS 6L/6R panel, but no background in lane 10.
        There is nothing that looks like a real defect in the panels.

        Figure 5K. Odd, indented appearance of bands. Little evidence of background.

      2. I see what you mean. I think Fig 7B looks a bit strange (and I wonder if you agree?). The 2 BMPR1A bands are the same distance from each other as the 2 GAPDH bands, but the BMPR1A bands are smaller. If they were loaded at the same time then surely the center point of each band should correspond to that of the one below (or could be mentally rejigged to approximate that if the 2 images got cut a bit out, as they have been here). I can’t see them lining up like that here. The noise levels between the 2 images are different as well, which is unlikely if they came from the same original image

        1. In reply to Erp May 11, 2013 at 7:30 am

          RE: Dev Biol. 2012 May 1;365(1):91-100

          You could be right. It is not quite obvious what was done. It is possible that the reactions in the panelsin figure 7B were run on different gels. I mentioned figure 7A top panel because it is all in one image.

        2. I mean the Dev Bio 2012 paper figure 7, you added the update whilst I was working it out, you are finding these problem papers too fast Fernando Pessoa 😉

          1. In reply to Erp May 11, 2013 at 8:20 am

            I found the irregularities a while ago and made a note of them.

            Nat Med. 2008 Sep;14(9):973-8. doi: 10.1038/nm.1852. Epub 2008 Jul 27.
            PlGF-MMP-9-expressing cells restore microcirculation and efficacy of cell therapy in aged dystrophic muscle.
            Gargioli C, Coletta M, De Grandis F, Cannata SM, Cossu G.
            San Raffaele Biomedical Park Foundation of Rome, 100 Via Castel Romano, 10028 Rome, Italy.
            PMID: 18660817

            Figure 1g. I doubt that the actin panel is the loading control for the ColI or the CD11b panels.
            I know that different mobility proteins may run as different arcs, but there is no reflection of the course of the bands in the Actin panel in the ColI or CD11b panels.

            Figure 2a. The bands in the Actin panel look like flattened out versions of the bands in the Actin panel of figure 1g. The course of the bands looks flattened out too. I suspect data reuse and manipulation.
            From Legends. Source of samples in lanes 4, 5 and 7 are not the same in figures 1g and 2a.

            Figure S2e. There is no background to anchor the bands.

            Figure S3a. MMP9 panel. There is a vertical darker blue line between the 1st and 2nd lanes (starting from the left).Suspect splicing.

            Figure S3b. There is no background to anchor the bands.

  11. Nature. 2006 Nov 30;444(7119):574-9. Epub 2006 Nov 15.
    Mesoangioblast stem cells ameliorate muscle function in dystrophic dogs.
    Sampaolesi M, Blot S, D’Antona G, Granger N, Tonlorenzi R, Innocenzi A, Mognol P, Thibaud JL, Galvez BG, Barthélémy I, Perani L, Mantero S, Guttinger M, Pansarasa O, Rinaldi C, Cusella De Angelis MG, Torrente Y, Bordignon C, Bottinelli R, Cossu G.
    San Raffaele Scientific Institute, Università Vita e Salute, Stem Cell Research Institute, Via Olgettina 58, 20132 Milan, Italy.
    Erratum in
    Nature. 2013 Feb 28;494(7438):506.

    This is what I first thought about this paper. Sorry that it is long, but it is to make points.

    Figure 4b.

    I think that there has been a splicing event between the 4th and fifth lanes of the right beta-SG panel.

    Notice the white vertical line separating the lanes and the two thin, grey bands to the right of this white line compared with the thick black band to the left of the lane.

    I think that the same MyHC panel has been used in the lower left MyHC panel and both MyHC panels on the right.

    In the right MyHC panels the right tip of the right-most line of continuous bands on the left has been cut off.

    In the left MyHC panel the left-most section (lane) of the MyHC panel has been cut off.

    If you put your finger over the “band” in the left-most lane of the right MyHC panels you will see how similar they are.

    Because the left-most “band” in the right MyHC panels goes up to the left you think they are different.

    I thought that a lot of the rest of the paper looked like that 3-D film that came out a few years ago. I only saw the trailer.

    Figure S3.

    The panels look more similar than different. a’ does not look like it has been processed in the same way as b’ and c’. b’ and c’ look quite similar. c’ has got more connective tissue, the island of white in the upper right quadrant. You can see a similar thing in b’ where you see the edge in the lower right corner. Possibly evasive.

    Then I read the authors’ reply to a comment on the original paper. The authors’ reply is the discussion E23-5.
    Nature. 2007 Dec 20;450(7173): discussion E23-5.
    PMID: 18097347

    Simpler explanation of results in discussion section E23-5.

    On enlarging figure 2d to 200% is that you can see a band at about the 200 level (marked by the arrow to the left of the panel) in the VAL TC lane (5th lane) and a very faint band at the 200 level in the VAR TC lane (6th lane). These bands are at about the same level as the left end of the band in the K1PR8 TC (left-most) lane, but above the level of the right end of this band, and even more above the level of the band in the N8PV5 TC (second) lane. If one assumes that the VLA BF lane is on the same blot as the left most two lanes this means that the projected position of the band at about 200 would be even lower than the band in the N8PV5 TC lane. This would easily explain why the band in the VLA BF lane (there is only one band in this lane, which is about a fifth down from the top of the panel, is lower down that the two “hat-shaped” bands in the VAL TC and VAR TC lanes.

    One distinct possiblity is that the VAL TC and VAR TC lanes and bands are from a different blot than the K1PR8 TC, N8V5 TC and VLA BF lanes. There is evidence already discussed that panels 1c and 1d are not from the same blot. So why should this be limited only to between panels? Why not within panels?

    My initial comments on figure 1b,c and d.
    The bars along the top are confusing.
    The molecular weight markers are drawn in at the side of the panels.
    In the top panel the VLA BF (untreated) lane does have a rectangle of signal where the wild-type is indicated at the side.
    If we were to take the lower panel (Ponceau S) at face value there is about ten to fifteen times less protein in the VLA BF lane than in the two lanes to the left of it and about 3 times less protein in the two lanes to the right of it.
    MANEX7B panel.
    There is are two thin vertical lines between the 1st and 2nd lanes (starting from the left). The bands in the rest of the panel look like they are part of a piece, but it is not good practice.
    The first two lanes are overexposed. You can only see one band. Overexposure so you only get one band is not proof that there were two bands.
    The bands in the 4th and 5th lanes look like hats, you cannot say that there are two bands.
    There is no signal in the 3rd lane.
    DYS2 panel.
    First two lanes, I can see only one band. The third lane does have a band.
    The 4th and 5th lanes each have a single band. These bands are not overexposed.
    The band in the third lane does migrate marginally faster than the bands in the 4th and 5th lanes, but
    it is not obvious that the bands are migrating along a front.
    If the authors want to lay great stress on the marginally faster band they need to demonstrate this rigorously, which they have not.
    The thin bands which appear a bit more than halfway down the MANEX7B and DYS2 panels give a clue to the orientation of the bands. In the MANEX7B panel these bands are sloping gently up to the right . In the DYS2 panel they are sloping down to the right. The bands have similar motility (it does not matter if they are the same thing) so should be highly comparable for what has been going on to molecules with their motility.
    My conclusion is that the MANEX7B and DYS2 panels are not from the same blot.
    My repetition.
    In the MANEX7B panel it looks like the bands are migrating along a front,
    whereas in the DYS2 panel it is not obvious that the bands are migrating along a front.
    If you put your finger over the band in the 5th lane of the DYS2 panel then you can easily imagine (which is what the authors ask the viewer to do all the time, for example in Nature. 2006 Nov 30;444(7119):574-9 they shifted the beta-actin panel over by one lane so you thought that things were different) that the band in the 3rd lane is about the same molecular weight as the band in the 4th lane.
    I think that there is no evidence that the band in the 3rd (VLA BF, untreated) lane is of lower molecular weight than the bands in the 4th and 5th (treated lanes).
    Even if it were of a lower molecular weight protein it might have activity.
    The fact that you don’t see a band in the 3rd lane in the MANEX7B panel is that this panel is likely from another blot. It and could be of anything.
    Poor quality gels in themselves are not misconduct, but they are not evidence of what they say.
    The authors play on the misunderstanding most people have.
    “The exception proves the rule” means the exception tests the rule. In German “probieren” still means to test, or “probe” in English.
    It does not mean that exceptions validate a rule or that exceptions can be explained away.
    The onus is on the authors to support what they claim.

    1. Re: Nature. 2006 Nov 30;444(7119):574-9.

      Again, people do not like it when odd things are pointed out.

      Nature must have found some merits in my arguments as a correction was issued and Nature is considering the unnatural aspects found in the correction. For example, the near straight and vertical right edge of the band in the 7th lane of the MyHc panel, and the less pronounced, but still odd, near vertical and straight left edge of the band in the 3rd lane of the MyHC panel.


      1. In the *corrected* pdf, if you compare the Varus and Vampire MyHC lanes at high magnification and increase the contrast a bit, you will find the same speckle above the left-but-one/left band. This further supports the suggestion that the two panels are the same (shifted by one and with the contrast slightly altered).

        1. In reply to simpliste May 15, 2013 at 3:04 pm

          ” if you compare the Varus and Vampire MyHC lanes at high magnification and increase the contrast a bit, you will find the same speckle above the left-but-one/left band”.

          Nicely spotted!

          Nature. 2006 Nov 30;444(7119):574-9

          Original image of figure 4b Nature. 2006 Nov 30;444(7119):574-9,

          which it now seems has been removed from the electronic version.


          I imagine it will be still be in the printed version, but they are so small in Nature.

          1. I’m not normally a conspiracy-theorist, but…there’s something else weird about the Nature correction (apologies if this was discussed already)…check out the b-SG blot shown in the “Corrigendum” (http://www.nature.com/nature/journal/vaop/ncurrent/full/nature11976.html) and compare it to the same blot in Figure 4b of the most recent version of the paper (http://www.nature.com/nature/journal/v444/n7119/full/nature05282.html). They’re not the same! The version in the corrigendum has clearly been “smoothed” Photoshop-style, hiding the blot-splicing lines that are apparent in the version that’s in the paper.

  12. Development. 1999 Oct;126(19):4247-55.
    Transplacental delivery of the Wnt antagonist Frzb1 inhibits development of caudal paraxial mesoderm and skeletal myogenesis in mouse embryos.
    Borello U, Coletta M, Tajbakhsh S, Leyns L, De Robertis EM, Buckingham M, Cossu G.
    Istituto Pasteur-Cenci Bolognetti, Dipartimento di Istologia ed Embriologia Medica, Università di Roma ‘La Sapienza’, Via A. Scarpa 14, Italy.
    PMID: 10477293


    Figure 3C.

    Is there a band of the indicated size in lane 2 or not? I am not sure.

    I know that stronger bands will be bigger than weaker ones and may even merge with bands in adjacent lanes. What I don’t understand is why the bands in lanes 1 and 5 are wider than the bands in lanes 2 to 4, yet the gap between the bands in 4 and 5 is not narrower than the gap between the bands in lanes 2 to 4, and the gap between the bands in lane 1 and 2 is wider than the gaps between the bands in lanes 2 to 5.
    Stronger bands don’tmake lanes wider.

    Legend figure 3C.
    (C) Embryos from injected and
    control placentas were isolated 24 hours after the injection of WOP
    cells previously transfected with an expression vector expressing a
    tagged (HA) Frzb1. Pools of eight injected and control embryos were
    lysed in RIPA buffer and immunoprecipitated with an anti-HA
    antibody; the immunoprecipitate was separated on SDS-PAGE,
    transferred onto nitrocellulose and blotted with the same antibody. A
    band corresponding to the expected molecular size (arrow) can be
    detected in the supernatant of transfected WOP cells (lane 1) and in
    samples of Frzb1-injected embryos after 24 hours (lane 3) but not in
    the supernatant of control WOP cells (lane 2) nor in samples from
    mock-injected embryos (lane 4). 48 hours after injection a faint band
    can still be detected in samples from injected embryos (lane 5).

    1. Not a nice figure is it? Lane 2 also has sharp enough edges to perhaps indicate splicing. Would love to see the “original gel”.

      But! But! Can that be THE Eddy De Robertis and THE Margaret Buckingham nested amongst the authors? Big names in developmental biology just don’t get any bigger. These people have real achievements to their names that have already stood the test of time.

        1. Then I see two opposing possibilities. Should we follow Euripides: “Judge a man by the company he keeps”. Or should we allow for a second, more modern, possibility: “I’m a developmental biologist… Get me out of here! ”

          Cossu did time in the Pasteur in the 90s and they published – actively and apparently enthusiastically – together, thus placing Euripides to the fore.

          Against that, the authorial contributions from the Stem Cells paper might provide an important clue:

          “Author contributions: G.M.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing; D.S.: conception and design, collection and assembly of data, data analysis and interpretation; G.M. and D.S. contributed equally to this work; S.M.: collection of data; B.G.: collection of data; E.S.: collection of data; R.T.: provision of study material; L.D.A.: collection of data; S.B.: collection of data; F.R.: provision of study material; M.B.: conception and design, manuscript writing, financial support; G.C.: conception and design, manuscript writing, financial support.”

          Of the authors, only G.M. and D.S. admit to “assembly of the data”. I interpret this statement that they – and only they – composed the published figures. Given that “G.M. and D.S. contributed equally to this work” – did they assemble the fictitious Fig. 1A equally and together? We here all know that no author of a paper is free from responsibility so that, if they can’t be bothered to read the thing, they fully deserve the embarrassment: Nevertheless we should not forget that the figure fabricator is the prime culprit.

          In the absence of further enlightenment from the tight-lipped and minimalistic UCL who, if they did their job properly, would know which author assembled each figure (and from whose primary data) in the fingered papers, we don’t know how to apportion blame – except from that contributions statement. For the moment, therefore, I continue to favour the second possibility: “I’m a developmental biologist… Get me out of here! “

          1. In reply to Scrutineer May 14, 2013 at 2:37 pm

            I really like you analysis. My earlier comment was a question.
            fernando pessoa May 14, 2013 at 1:36 am

            You have helped answer the question.

            The reason I asked the question was that I didn’t fully understand what you meant.
            Scrutineer May 12, 2013 at 1:36 pm
            Of figure 3C Development. 1999 Oct;126(19):4247-55
            you commented “Not a nice figure is it? Lane 2 also has sharp enough edges to perhaps indicate splicing. Would love to see the “original gel”.”

            Then you mentioned that 2 of the other authors are big names in developmental biology.
            I took this as an possible example of Euripides: “Judge a man by the company he keeps”.
            I wasn’t sure, hence the question. Now I see you meant “I’m a developmental biologist… Get me out of here! “

          2. Fernando, while on the Euripidean topic of the company a man keeps, as befitting a rising star of translational bioscience, Cossu is on a bunch of editorial boards, most notably being a senior and founding editor of EMBO Molecular Medicine which:

            “Publishes high quality research relevant to the understanding, prevention and treatment of human disease.”

            Oddly, though, he doesn’t publish his own work there. Is there anything in the above statement that would prevent him from doing so?

  13. Circ Res. 2010 Apr 16;106(7):1290-302. doi: 10.1161/CIRCRESAHA.109.206045. Epub 2010 Feb 25.
    Sox2 transduction enhances cardiovascular repair capacity of blood-derived mesoangioblasts.
    Koyanagi M, Iwasaki M, Rupp S, Tedesco FS, Yoon CH, Boeckel JN, Trauth J, Schütz C, Ohtani K, Goetz R, Iekushi K, Bushoven P, Momma S, Mummery C, Passier R, Henschler R, Akintuerk H, Schranz D, Urbich C, Galvez BG, Cossu G, Zeiher AM, Dimmeler S.
    Institute of Cardiovascular Regeneration, Centre for Molecular Medicine, University of Frankfurt, Theodor-Stern-Kai 7, 60590 Frankfurt, Germany.
    PMID: 20185800

    Figure 1.
    Figure 1F. The bands in the CD131, CD73 and KDR panels are uncannily similar.
    There is no background between the bands (no background is equivalent to splicing between each lane).
    Figure 2.
    Figure 2D. Fli-1 panel. There is no background between the bands.
    The bands in lanes 1 and 2 (starting from the left) look very similar. The bands in lanes 3 and 4 look very similar.
    The band in lane 5 looks like a longer exposure of the bands in lanes 3 and 4.
    GAPDH panel. All the bands look very similar.
    Figure 2G. Islet 1 panel. No background between bands.
    The bands in lanes 1 and 2 look very similar.
    The bands in lanes 3 and 4 look very similar. Possibly mirror images of each other.
    GATA4 panel. The bands in lanes 2 to 4 might be different exposures of the same thing.
    TBx5 panel. The bands in lanes 1, 2, 4 and 6 look very similar.
    Mef2C panel. The bands in lanes 4 and 6 look very similar.
    GAPDH panel. I suspect that the bands are different exposures and re-sizing of the same original image
    Figure 3.
    There is no background between the bands.
    Figure 3C, right panel. The 3 error bars are indistinguishable.
    Figure 3G. h-Tie2 panel. The bands in the human heart and cMAB panels look like they could be images of the same thing.
    The band in the human heart lane of the h-GAPDH panel looks very similar to the band in the MI cMAB lane of the GAPDH panel.
    GAPDH panel. The bands in the human heart and cMAB lanes look very similar. The bands in the mouse heart, MI PBS and MI cMAD lanes look very similar.
    Figure 4.
    Figure 4A.
    There is no background between the bands.
    The band in the nanog panel (mES cell lane) looks very like the bands in the GAPDH panel.
    The band in the mES cell lane of the Oct3/4 panel and the bands in lanes 1 , 3 and 4 of the Sox17 panel look similar.
    I suspect that they are different exposures of the same thing. I suspect that thy will turn out to be an longer exposure of the band in the nanog panel.
    Figure 5.
    Figure 5A. There is no background between the bands. The bands in the GAPDH panel look very similar to each other, especially the first and last bands.
    Figure 5B and 5C. The error bars are indistinguishable.

  14. Development. 2002 Jun;129(11):2773-83.
    The meso-angioblast: a multipotent, self-renewing cell that originates from the dorsal aorta and differentiates into most mesodermal tissues.
    Minasi MG, Riminucci M, De Angelis L, Borello U, Berarducci B, Innocenzi A, Caprioli A, Sirabella D, Baiocchi M, De Maria R, Boratto R, Jaffredo T, Broccoli V, Bianco P, Cossu G.
    Stem Cell Research Institute, Dibit, H. S. Raffaele, Via Olgettina 58, 20132 Milano, Italy.
    PMID: 12015303


    Figure 5.

    Figures 5B and 5E. There is no background joining most of the bands. The PCR bands have an
    unnatural look about them, some with close cropping around them.

  15. J Cell Biol. 2002 Aug 19;158(4):731-40. Epub 2002 Aug 19.
    REN: a novel, developmentally regulated gene that promotes neural cell differentiation.
    Gallo R, Zazzeroni F, Alesse E, Mincione C, Borello U, Buanne P, D’Eugenio R, Mackay AR, Argenti B, Gradini R, Russo MA, Maroder M, Cossu G, Frati L, Screpanti I, Gulino A.
    Department of Experimental Medicine, University of L’Aquila, 67100 L’Aquila, Italy.
    PMID: 12186855

    Figure 1.


    Figure 1B. REN panel. Suspect that the right-most lane (4) has been spliced on. There is a thin white, vertical streak between lanes 3 and 4, and the background has a different quality

    Figure 1C. GAPDH panel. I suspect that the bands in lanes 2 and 5 (starting on the left) are likely the same. The 2 dots above the right end of the band in lane 5 distract the eye.

    I suspect that all the bands in the panel are versions of each other. The lower parts have likely been planed off in some. For example, in lane 2 the band has a flatter bottom than in lane 1.

    Figure 1D. PC12 set of panels. REN panel. Suspect that lane 2 have been spliced in. Vertical changes in background between the lanes 1,2 and 3.

  16. Proc Natl Acad Sci U S A. 2006 Nov 7;103(45):16995-7000. Epub 2006 Oct 31.
    MyoD expression restores defective myogenic differentiation of human mesoangioblasts from inclusion-body myositis muscle.
    Morosetti R, Mirabella M, Gliubizzi C, Broccolini A, De Angelis L, Tagliafico E, Sampaolesi M, Gidaro T, Papacci M, Roncaglia E, Rutella S, Ferrari S, Tonali PA, Ricci E, Cossu G.
    Department of Neurosciences and Interdisciplinary Laboratory for Stem Cell Research and Cellular Therapy, Catholic University, Largo A. Gemelli 8, 00168 Rome, Italy.
    PMID: 17077152

    Figure 3.


    Figure 3B. Left (PCR) panels.
    MyoD panel. I don’t understand why the left lane has signal from floor to roof, yet the right lane has less signal in the middle, where you might expect a band.

    Figure 3C. Phosho-p38 panel. Vertical, grey streak between lanes 6 and 7.
    Beta-actin panel. The background is smooth and grey.

    Figure 5.


    Figure 5. Western panels.

    Myosin panel. The background is smooth and grey. This means that you cannot pin down the bands.
    Lower beta-actin panel. The background is smooth and grey. This means that you cannot pin down the bands.

    Figure 6.


    Figure 6C. Western panels.
    Myosin panel. Suspect splicing between all 3 lanes. Vertical streaks between the lanes.
    Beta-actin panel. Suspect splicing between the left and right lanes. Change in background between the lanes. Also the band in the left lane is black, whereas the bands in the middle and right lanes which look saturated are only grey.

    1. Please, please fernando, could you post your comments on these papers in PubPeer?

      They are high standard queries, and that excellent initiative is just in the need for such material. Actually I think with time they will have deeper effect in prompting a reaction from authors and reasoning upon readers while posted there than anywhere here.

        1. Not sure why anyone would consider pubpeer the corporate state??? I also think it would be great if fernando would comment on pubpeer.

  17. As way of explanation. There is a group of investigators associated with G Cossu in the Optistem network.

    The publication below struck me as odd.

    EMBO Mol Med. 2012 Sep;4(9):910-23. doi: 10.1002/emmm.201101075. Epub 2012 Aug 23.

    Loss of a single allele for Ku80 leads to progenitor dysfunction and accelerated aging in skeletal muscle.
    Didier N, Hourdé C, Amthor H, Marazzi G, Sassoon D.

    Myology Group, UMR S INSERM, Université Pierre et Marie Curie Paris VI, Pitié-Salpétrière, Paris, France.


    Figure 1D. Where are the 18 month +7- and -/- panels? Is the magnification the same for all panels? It is difficult to see the scale bar in the left 2 panels.

    Figure 2A. Where are the 18 month +7- and -/- panels? The right-most panel (-/-) looks very much like the second panel (+/+). If you simply cover over the vertical right-most strip of the right-most ( -/-) panel you will see what I mean. Muscle is variable. You can see this in the panel the authors have selected.

    Figure 2B. None of the panels looks that different except for the orange stain in the middle and also in the right panel. The large empty roundish spaces clustered mostly in the lower left of the right-most (-/-) panels might be adiposites. Wouldn’t you expect the stain to be specific? Much of the orange is present in cells which I take as muscle fibres, not adipocytes.”
    After showing this figure to some histopathologists they said that all the empty roundish spaces were adipocytes, which should all have stained brightly orange with Oil Red O. The opinion was that the orange shown in figure 2B was spilled on the sections and not properly washed, i.e. artefact.

    Figure 2C. Where are the 18 month +/- and -/- panels?

    Figure 5G. Is there a band in the right-most band of the p21 lane? I can see the left end of a band, and a slight impression of the rest of a band. Is the result due to a poor blot?

    Figure 5H. The band in left-most lane of the PW1 panel is obscured with white flecks.

    Figure 6C. As we can see from the left panels the picture is highly variable. Each of the left panels has an area which is normal. The right panels could simply be due to sampling.

    Figure 7E. It is true that modified and unmodified bands should be superimposable.
    This does not mean the same. The total PTEN bands could simply be a longer exposure of the p-TEN (Ser380/Thr382/383) bands.
    Note that the same dark schmutz below the left end of the band in the middle lanes appears in both panels,
    as do other smudges in the first lanes (1/3rd in from right end of the bands in the first lanes at the bottom of the panels), also there is smudge 1/3rd in from right ends of bands in the right-most lanes of both panels.

  18. It is really annoying to see how NATURE that has gained respect for its very high publication barrier fails in removing fake reports. When reading Nature we find corrections but almost never retractions, even when evidence against a published paper is VERY STRONG as in this case. Regrettfully, thanks to Retraction Watch, we are seeing that data fabrication is not so unlikely, so the very few retractions posted by NATURE could not pass any statistical test.

    There are millions of dollars spent by funding agencies, huge industrial interests and lots of postdoctoral careers burned because of chasing down the wrong premises of false reports. I feel that NATURE and the other heavy league journals should feel at least in part responsible for this incredible waste of people, money and resources. Scientists and Investors trust these journals and have faith in their judgment. If you buy a vehicle from a car dealership because of the enthusiastics comments of the salespeople and after a few weeks the car blows up, who is to be considered responsible for this failure? The factory yes, but also the car dealership because by contract their technicians should have checked that the vehicle was indeed in good shape, the engine running, wheels are not loose, the fuel not leaking, and later process warranty claims.

    A jury might convict NATURE of misconduct if they keep not informing their clients of potential problems they are aware of. J Cell Biology does this very effectively. The world science business has grown to a point that it is now impossible to ignore the social and economical damage caused by data fabrication. Please invest some money, one serious investigative journalist and one laptop computer might be enough for NATURE, don’t be cheap, be wise and adopt the J Cell Biology policy before it is too late!

    1. Science. 2003 Jul 25;301(5632):487-92. Epub 2003 Jul 10.

      Cell therapy of alpha-sarcoglycan null dystrophic mice through intra-arterial delivery of mesoangioblasts.
      Sampaolesi M, Torrente Y, Innocenzi A, Tonlorenzi R, D’Antona G, Pellegrino MA, Barresi R,Bresolin N, De Angelis MG, Campbell KP, Bottinelli R, Cossu G.

      Stem Cell Research Institute, H. S. Raffaele, Via Olgettina 58, 20132 Milan, Italy.

      Figure 1.


      Figure 1 4D.

      Int beta 1 panel. Low resolution. Difficult to tell what is a band,and what is not a band. There seems to be a constant thick black band, and then in the Gas 30 and 60 lanes a lower black band appears. Which is the band of interest?

      MhHC panel. The background is monotonous grey. This means that you cannot pin down the bands. Even if bands are touching does not mean that they belong together.

      alpha-SG panel. The bands are on a strip which is narrower than the panel. There are horizontal changes in background above and below the bands.

      Figure 4.


      Figure 4A. I think that the tartan-like appearance is not true background.


      Supplementatry figure 1D. The figure is small.

      LacZ panel. If you click twice (the resolution is still good) with the internal control which comes wth the official website you will see that the band in the C lane has a near vertical, straight right edge and there is a dark area to the right of that.

      The background in lane C to Sol does not continue into the M lane. To the left of the bands in the M lane the background is black.

      To see the background it helps to bring the computer screen upright.

    1. It is a bit worrying that in this 13th May 2013 correction of J Cell Biol. 2004 Jul 5;166(1):97-109.

      I cannot find the original figures.


      In particular it would be useful to be able to see the original figure 9B.

      “In addition, the original version of Fig. 9 B showed a composite panel for Total Erk1/Erk2 that contained three duplicated lanes. The authors have indicated that the wrong panel was inserted due to a clerical error during figure preparation. A corrected version of Fig. 9 B is shown below.”

      How does inserting the wrong panel result in three dupplicated lanes?

  19. In reply to Scrutineer May 15, 2013 at 4:00 am

    I don’t know if it odd that G Cossu doesn’t publish primary experimental work in EMBO Molecular Medicine. You don’t have to publish primary experimental work in the journal of which are an editor. Many think it better if you don’t.

    With reference to:

    “Publishes high quality research relevant to the understanding, prevention and treatment of human disease.”

    where is the evidence?

    G Cossu has published primary experimental work in Skeletal Muscle

    where he is on the editorial board.

    as are Margaret Buckingham, David Sassoon, Pura Muñoz-Cánoves, Shahragim Tajbakhsh, Peter Zammit, and Emilio Clementi, who are his associates


    In addition the editorial board member Vittorio Sartorelli has published with G Cossu.


  20. Proc Natl Acad Sci U S A. 2004 Nov 23;101(47):16507-12.

    Mitochondrial biogenesis by NO yields functionally active mitochondria in mammals.
    Nisoli E, Falcone S, Tonello C, Cozzi V, Palomba L, Fiorani M, Pisconti A, Brunelli S, Cardile A, Francolini M, Cantoni O, Carruba MO, Moncada S, Clementi E.

    Author Affiliations
    *Department of Preclinical Sciences, Laboratorio Interdisciplinare Technologie Avanzate (LITA) Vialba, University of Milan, 20157 Milan, Italy; †Istituto Auxologico Italiano, 20149 Milan, Italy; §Stem Cell Research Institute, Dipartimento di Biotecnologie, Ospedale San Raffaele, 20132 Milan, Italy; ¶Eugenio Medea Scientific Institute, 23842 Bosisio Parini, Italy; ∥Istituto di Farmacologia e Farmacognosia, and **Istituto di Chimica Biologica, University of Urbino “Carlo Bo,” 61029 Urbino, Italy; ††The Wolfson Institute for Biomedical Research, University College London, London WC1E 6BT, United Kingdom; and ‡‡Department of Pharmaco-Biology, University of Calabria, 87036 Rende, Italy

    Department of Preclinical Sciences, Laboratorio Interdisciplinare Technologie Avanzate (LITA) Vialba, University of Milan, 20157 Milan.

    Splicing dilutes meaning.

    Figure 2.
    PCR panels very small even at highest magification of official webpage (an issue in itself).
    U937 panel. Difficult to spot, but there are thin, grey, vertical lines between the C and DETA-NO and between the DETA-NO+ODQ and 8 Br-cGMP lanes. Suspect splicing.
    L6 panel. Difficult to spot, but there are thin, grey, vertical lines between the C and DETA-NO, between the DETA-NO+ODQ and 8 Br-cGMP lanes, and between the 8 Br-cCGMP and BAY 41-2272 lanes.
    Suspect splicing.
    PC12 panel. Difficult to spot, but there are thin, grey, vertical lines between the C and DETA-NO, between the DETA-NO+ODQ and 8 Br-cGMP lanes, and between the 8 Br-cCGMP and BAY 41-2272 lanes.
    Suspect splicing.

  21. Science. 2005 Oct 14;310(5746):314-7.
    Calorie restriction promotes mitochondrial biogenesis by inducing the expression of eNOS.
    Nisoli E, Tonello C, Cardile A, Cozzi V, Bracale R, Tedesco L, Falcone S, Valerio A, Cantoni O, Clementi E, Moncada S, Carruba MO.
    Integrated Laboratories Network, Department of Preclinical Sciences, Luigi Sacco Hospital, Milan University, 20157 Milan, Italy.

    There is frequent splicing in the panels you can see.
    It does dilute the meaning.

    Figure 1.

    Figure 1A. Straight, vertical discontinuities in background between the mtDNA and left AL lane, and between the the right AL lane and the left CR lane. Suspect splicing.

    Figure 1D. The right edge of the background between the bands in the M lane is straight and vertical. Suspect splicing.

    Figure 2.

    PCR panels for brain, liver, heart and brown adipose tissue are very small even at the largest maginfication of the official webpage.

    The liver and heart panels are the same.

    Liver panel. There is a vertical discontinuity inbackground between the AL and CR lanes. Suspect splicing.
    Heart panel. There is a vertical discontinuity inbackground between the AL and CR lanes. Suspect splicing.

    Brown adipose tissue. There is a thin, vertical white streak between the tops of the M and AL. Possible splicing.


    Figure S1. The AL and CR lines are smooth. All the symbols lie on the the lines.
    The error bars you can see are mostly very similar.

    Figure S2. The 4 PCR panels are very small.

    The brain and heart panels look very similar.
    The AL and CR lanes in the heart and brown adipose panels look very similar.

    Brown adipose tissue panel. The left edge of the trail in the AL lane is straight and vertical. Suspect splicing.

    1. fernando,
      why not put these directly on pubpeer? there community seems to be getting pretty active and authors are starting to reply.

      1. It is not possible. Hadn’t the people who run Pubpeer thought about that possibility instead of keeping on nagging with the same request? It is Big Brother wanting to know you institutional address. You never whose hands the information will fall into.

        Feel free to cut and paste the comments into pubpeer.

        1. Sorry Fernando, but could you please elaborate as this topic very much interests me? Could you please clarify why you think Pubpeer a Big Brother?

          I became aware of its existence quite recently, as it was recommended by Paul B. of Science-Fraud.org. I find the idea fantastic, and I am not sure why it did not soar. I was a reader of SF and I remember reading many comments from you there. Why not trust Pubpeer and trust RW? Should I also suspect Pubpeer?
          I think your interesting queries on RW will soon be forgotten in a pile of comments in a few days, and possibly completely lost after few years.

          Just note that anyone without an account can comment on Pubpeer. And I am using my institutional email right now.

          1. Plus — not my question related but a relevant observation — but I do think some people from your sphere are probably fairly sure who you (or Claire Francis) are, as you apparently stick to one or few alias and fields of expertise, and maybe one would not exactly need such elaborate schemes to guess an institutional email in this case.

          2. CR wrote “Could you please clarify why you think Pubpeer a Big Brother?”

            Some whistleblowers need anonymity. That’s for their own protection. The science-fraud site goings on reported here is a case in point.

            Whistleblowers also need to know the sites they send suspect data to can be trusted.

            I somehow doubt Fernandos writings will be forgotten any time soon.

          3. Dear Stewart — thanks, but please have a look into Pubpeer.com. It seems to offer anonymous commenting even more than here, where an email has to be provided. One does not have to log in. Actually this seems to be the niche of the initiative.

            “I somehow doubt Fernandos writings will be forgotten any time soon.” — Maybe in your personal memory, but they will be very hard to find by any passing readers quite soon.

  22. We’d like to clear up a few issues about PubPeer.com

    CR is correct. At PubPeer there are two ways of posting comments: Either by creating an account using an institutional email address to leaving a comment after signing in (either anonymously or signed) or by submitting a comment to the system without signing in. Wheras comments submitted after signing in appear on the site immediately, the submitted comments are screened for spam and for various other issues outlined in the site’s FAQs. The only way for us or anyone to track the origin of a comment submitted using this second option is through the IP address of the computer used to submit the comment. The IP address of a computer is very easily obfuscated (see the torr project for instance). We at PubPeer take anonymity very seriously and that is why we recently added this second option for submitting comments.

    As for us being a “corporate state”, fernando is correct. The domain name PubPeer.com is owned by an LLC. However, we have set this up for the sole purpose of attempting to protect our families and private bank accounts from any litigation brought about by people like this guy: http://www.retractionwatch.com/2013/05/03/two-expressions-of-concern-in-blood-for-md-andersons-aggarwal-who-has-threatened-to-sue-retraction-watch/. We pay for all of the server fees and LLC fees out of our own pockets and do not accept any outside money (via advertising or any other form).

    1. Thanks for clearing that up. Also, lemme know if you guys do a thread on post-pub commenting. I know some folks in really little fields that fear to say anything at all, lest they be IDed even with anon commenting. I feel I’ve been getting annoyed about this for the past 8 years, and things are reaching a breaking point. And I got the email records to show it, heh.

    2. I very much encourage Pubpeer, thanks for clarifying. I yet feel some who indeed like the thrill of vigilantism do not quite appreciate the idea of true anonymity, as they have to drop their aliases and actually their connection with other previous cases. They will not anymore get threads about them like “Claire Francis scores a bulls eye” etc, if they manage retractions via Pubpeer.
      However I feel this way it is much more formal and sane.

      1. CR qrote “….pubpeer…..I feel this way it is much more formal and sane”

        Are there any retractions that have occured via pub peer?

        Any at all?

        I must say i think Francis has done alot more for science than you yet realise.

        I hope, in years to come, we all know who Francis is….but not until we know the career of Francis, and others, will not be obiterated by those who wish to redo the science-fraud saga.

        Rather than suggesting a lack of sanity I would suggest a Francis brings a much-needed breathe of fresh air into the scientific arena.

        1. Stewart, sorry: I do realise the benefit of Claire and awfully many other reviewers to Science. I emphasize these “many others” took no personal or “archetype-personal” credit for being good reviewers.
          What they do is merely Post-Publication Peer Review (PPPR) and is a necessary, natural, mechanism of scientific progress. Naturally pppr results in retractions/corrigenda and often in political retaliation — couldnt be otherwise. This is why peer review is usually meant to be anonymous, although often we guess who is at the other end.
          Thus being, I think — and I mean no offence by those–,: i) that getting too passionate/personal about a mechanism of peer review doesnt seem sane or healthy, for both sides of the issue; ii) that some PPPR agents are often called “vigilants” just because of their posture (e.g. attracting much public attention, jesting, sticking to codenames, etc) which indeed resembles comic strip characters; iii) that naturally many people in their fields know quite well who is Claire and other several active PPPR peers (i.e., this not being exactly a “the world must know” kindda thing), iv) Pubpeer.com is just one recent attempt to make PPPR safer, easier and more natural and is potentially more efficient and cleaner (though possibly less involving) than inflamed blogs and email aliases.

          PPPR mechanisms will inevitably evolve, and retractions will surely follow: let us just watch (and contribute!). I hope all current PPPR agents will adapt and keep up their good job.

          1. In reply to CR May 16, 2013 at 6:43 pm

            There is no “i” in my first name.

            There can be more than one way of post-publication review. Many have been doing post-publication review for many years. The journals, journalists, writers…blogs.

            I like this one, which I often find quite funny (deadpan):


            Very informative:


            Retraction Watch is part, perhaps the driving part (most hits), of post-publication review. I think it started in August 2010. Pubpeer is another part, but is by its nature more exclusive. Publication does mean something for the public to see. Retraction Watch fits that bill very well. The public pays for most of the research in most countries. In the U.S. directly through agencies such as the NIH, but also through tax-breaks for research organisations.

            What has “often in political retaliation” got to do with anything? If the work is real it will stand up by itself.
            I simply notice things odd on the page. If there were nothing odd on the page I would have nothing to say.

            When people send things into Retraction Watch there are often “Hat Tips” written at the bottom of the post.
            Don’t you see them?

            I don’t write the posts, or the titles of the posts. I think that Ivan and Adam have gots minds of their own.

            ” sticking to codenames”. Is CR a real name or a code? When things are ridiculous people will laugh. That is in the mind of the beholder. Please tell me which things you thing were jesting as they obviously made you laugh or smile.

          2. CR…so you don’t have a single example of a retraction by pubpeer reviews? I don’t doubt there will be any either as that is not the purpose of the site.

            I am sure pubpeer may well have a role for whatever is what devised for….but it is not suitable, in my opinion, for the intense science-fraud investigations we all need to move forward.

          3. Dear Claire — I really appreciate your efforts and I do not mean to diminish those in anything. Sorry if you felt included in some of my remarks, but actually I did specifically mean to. Also, I do not want to push topic off the track any longer, and I think I have made my points clear and they might make sense to some. Time may prove me right or wrong about the way things are changing, and Pubpeer.com is so young…
            Thanks for the suggested links!

          4. oops — “Sorry if you felt included in some of my remarks, but actually I did NOT specifically mean to.”

  23. In reply to Freeloader May 15, 2013 at 3:51 pm

    According to the G Cossu “after reading our reply, no one considered the issue
    any longer”.

    Don’t you get it that you are not supposed to question such people?

    I didn’t send G Cossu my concerns about EMBO Mol Med. 2012 Sep;4(9):910-23. doi: 10.1002/emmm.201101075. Epub 2012 Aug 23.

    because he is an assoicate of the senior author and hence conflicted.


    I thought that was the right thing to do, and that it would have been the correct thing for the journal to have kept my concerns from him.

    From: Giulio Cossu
    Date: Sat, Nov 10, 2012 at 8:25 AM
    Subject: Concerns 1; concerns 2
    To: clare.francis1946@googlemail.com
    Cc: Stefanie Dimmeler

    Dear Mrs. Francis,

    I do not know who you are and why you sent around to a number of people,
    except me, your accusations of 1. Conflict of interest and, 2. Fraud.

    As for 1, I am copying the Editor in chief of Embo Molecular Medicine who
    will confirm that, just because there could have been a conflict of
    interest, I did not see the ms by Sassoon, except than after it was

    As for 2, I wasted enough time with another mentally deranged person, Alan
    Bretag, to go into details of your accusations again. There was a matter
    arising in Nature, and after reading our reply, no one considered the issue
    any longer. I hope you may spend more fruitfully your time in the future.

    With best regards,

    Giulio Cossu

    Department of Cell and Developmental Biology,
    University College London
    Rm 545, Rockefeller Bldg. 21 University Street.
    London WC1E 6DE
    Tel: +442076796947
    Fax: +442076799349
    E-mail: g.cossu@ucl.ac.uk

    1. Keep on walking, because the machinery goes on (and the wasting of money).

      What do you do when you have a stone in your shoe? If it causes too much disturbance you probably will try to remove it.

      But what do you do if you are the stone?

  24. I’m not sure I would take Dr. Cossu’s word that no one is considering the issue any longer! What do you think about Nature publishing two versions of the corrected figure… one “polished”, larger version for the corrigendum and an unmodified (smaller) version in the corrected paper? It looks shady to me.

    1. In reply to freeloader May 15, 2013 at 7:26 pm

      “I’m not sure I would take Dr. Cossu’s word that no one is considering the issue any longer!” is exactly what I think too.

      There’s no statute of limitation on scientific discussion. Nature. 2006 Nov 30;444(7119):574-9 is not so long ago, and now we have Nature. 2013 Feb 28;494(7438):506, which is really recent and with its own set of problems.

      What suprises me is how they could find protein samples from 2006 to make Nature. 2013 Feb 28;494(7438):506, yet could not find their notes to support J Cell Biol. 2006 Jan 16;172(2):233-44.

      The comment in Nature. 2007 Dec 20;450(7173):E23; discussion E23-5 is also not so long ago.

      I don’t think that there is anything wrong about pointing out things that don’t fit, or coming up with a simpler explanation that fits the facts.

      Just because something sits in the literature for a few years does not make it true.

      Nature publishing two versions of the corrected figure is really weird. Was Nature given 2 versions, or did Nature klutz it up?

      1. this story stinks to heaven and is not doing to a favor to current european stem cell research. i’m just exasperated. what shall the EU think of this when being asked to fund further multimillion € framework programmes?

          1. hi stewart, i don’t know about any particular papers, but the optistem initiative is EU-funded. strategically speaking, i feel that this is a disaster to the muscle regeneration field. dimmeler has had her share of retractions, and now cossu is in the crosshairs. what were/are they thinking? all these papers have been published in the 2000s when publishing standards were active and implemented. stitching western blots or PCRs together was perhaps good for BBRC papers in the 80s, but not anymore…

            alright, you might tell me that people sitting on the EU FP study section groups serve themselves first, but still. it’s just very bad publicity, in the very sensitive stem cell field. the public is holding great expectations, but these days, the only things in the news are frauding stem cell clinics, human cloning and dodgy papers by eminent researchers. what a mess.

  25. Reply to plemplem May 18, 2013 at 1:40 am

    It might be a good idea to lance the boils.

    S Dimmeler has hundreads of papers. In many the bands have that very familiar moustache shape.

  26. Cell Death Dis. 2012 Nov 15;3:e418. doi: 10.1038/cddis.2012.159.

    Autophagy as a new therapeutic target in Duchenne muscular dystrophy.
    De Palma C, Morisi F, Cheli S, Pambianco S, Cappello V, Vezzoli M, Rovere-Querini P, Moggio M, Ripolone M, Francolini M, Sandri M, Clementi E.

    Department of Biomedical and Clinical Sciences, University Hospital L. Sacco, Università di Milano, Milan, Italy.

    Figure 2.

    Figure 2b. The band in the right lane of the mdx p62 panel and the band in the right lane of the WT p62 panel could easily be horizontal mirror images. You don ‘t get to see the right end of the band in the right lane of the mdx p62 panel.

    The bands in the left lanes of the actin panels are highly reminiscent.

    Figure 3.

    Figure 3a. I suspect that the bands in the left 2 lanes of the WT and mdx Actin panels started out the same.

    The band in the left lane of the mdx Actin panel has a pointy right end. I think this is unnatural.

    Figure 3a mdx LC3 I LC3 II 15hr + is inverted to become Fig 3b WT LC3 I LC3 II 15hr.

    Figure 3b. WT set of panels. Actin panel. I think that the bands in the Fed minus (left) lane and in the 15h + (right) lane are re-sized versions of the same original thing. Note the shape and the internal lighter areas.

    Figure 3c. Mdx set of panels. pAkt panel. The left end of the band in the right lane has a light halo. The area to the left of this is too bitght and clear. Suspect manipulation.

    Figure 7.

    Figure 7a. Controls LC3 I/LC3 II panel. Suspect splicing between the middle lanes. Step in level of bands.
    Controls p62 panel. Step in level of bands between middle lanes.
    DMD LC3 I/LC3 II panel. Suspect splicing between lanes 1 and 2. Light vertical streak between bands.

  27. J Cell Biol. 2006 Jul 17;174(2):231-43. Epub 2006 Jul 10.
    Complete repair of dystrophic skeletal muscle by mesoangioblasts with enhanced migration ability.
    Galvez BG1, Sampaolesi M, Brunelli S, Covarello D, Gavina M, Rossi B, Constantin G, Torrente Y, Cossu G.
    Author information
    1Stem Cell Research Institute, San Raffaele Hospital, 20132 Milan, and Department of Experimental Medicine, Human Anatomy Institute, University of Pavia, Italy.




  28. 2014 correction of 2013 Developmental Cell G Cossu paper.

    Dev Cell. 2013 Mar 25;24(6):586-99. doi: 10.1016/j.devcel.2013.01.022. Epub 2013 Mar 7.
    Dll4 and PDGF-BB convert committed skeletal myoblasts to pericytes without erasing their myogenic memory.
    Cappellari O1, Benedetti S, Innocenzi A, Tedesco FS, Moreno-Fortuny A, Ugarte G, Lampugnani MG, Messina G, Cossu G.
    Author information
    1Department of Cell and Developmental Biology and Centre for Stem Cells and Regenerative Medicine, University College London, WC1E 6DE London, UK.

    2014 correction.

    In the originally published version of this paper, Figures S4A–S4C correctly showed control cells; however, these same cells erroneously also appeared in Figures S4K–S4L. We have now replaced the latter three panels with the correct data from shDll4-treated cells and adjusted the panel letters to proceed alphabetically (revised Figures S4J–S4L). The corrected figure is shown below and has also been corrected in the online version of the article. The authors apologize for any confusion this error may have caused.

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.