Retraction Watch

Tracking retractions as a window into the scientific process

Nature corrects figures McGill committee found had been “intentionally contrived and falsified”

with 15 comments

nature 5 31The second of two corrections by McGill researcher Maya Saleh for what a university committee called “intentionally contrived and falsified” figures has run in Nature.

We reported in January that the McGill committee concluded that

two figures in [a] Nature paper had been “intentionally contrived and falsified.” One of those figures was duplicated in a PNAS paper, which also contained an image that  had incorrectly labeled some proteins.

The committee recommended corrections for both of the papers. The PNAS correction ran in February. Now, the Nature Corrigendum has appeared:

The immunoprecipitation lanes in Fig. 4c of this Letter were incorrectly derived from figure 6 of ref. 1 (which was generated prior to publication of this Letter). Also, some input lanes (“Total Casp12”) were incorrectly derived from figure 6 of ref. 1 and the anti-tubulin control right lanes of Fig. 4b were also inadvertently duplicated from the left lanes. Figure 1 of this Corrigendum shows the corrected Fig. 4b and c, representing new and independent experiments for Fig. 4c. We re-probed the original western blot in Fig. 4b with anti-tubulin. The experiment for Fig. 4c is robust and has been repeated a total of four times (twice each by two workers). The interpretation of the data and conclusions supported by it are unaffected; namely, that caspase-12 forms a complex with and co-immunoprecipitates with caspase-1 when co-expressed by transfection into human (HEK 293T) cells. There was also some association of caspase-12 with caspase-5 (more so than previously described) but very little with caspase-9. Our conclusions remain unaltered and the original legend for Figure 4 also remains correct.

nature12181-f1.2Figure 1: This figure shows the corrected Fig. 4c and the corrected anti-tubulin control panel of Fig. 4b of the original paper.

The correction has already drawn some puzzled comments from Retraction Watch readers. The paper has been cited 151 times, according to Thomson Scientific’s Web of Knowledge.

Although the McGill committee found issues with two other papers, they said they were due to “touchup of blemishes” and “artifacts created during the preparation of the scans for publication.”

Comments
  • Conrad T Seitz MD May 31, 2013 at 10:12 am

    You can class this with the “science was right but figures were falsified due to the rush of time, etc.” so now that we have the time, “here are the correct figures, and by the way we ran the protocol two more times just to be sure…” Since this paper has been cited over a hundred and fifty times, it must have some importance?
    I guess it’s OK, even though they fibbed, because it came out right in the end. Right?

    • Stewart May 31, 2013 at 11:22 am

      I love the line “Our conclusions remain unaltered” 🙂

      I’ve said it before and I’ll say it again – “we need diagrams with big red arrows”.

      More to see anyone?

    • michaelhbriggs May 31, 2013 at 12:50 pm

      My head is spinning.
      The Nature figure (4c 2006) had to be replaced because it had been incorrectly derived / intentionally contrived and falsified from figure 6 of PNAS (2008), but this figure (6, PNAS, 2008) was corrected because it was a mislabeled republication of figure 4c in Nature in 2006.

      So, in summary, the Nature figure was derived from the PNAS figure, which was a republication of the Nature figure, which was derived from the PNAS figure, which was a republication of the Nature figure, which was derived from the PNAS figure, which was a republication of the Nature figure, which was derived from the PNAS figure…

  • Freeloader May 31, 2013 at 12:03 pm

    In addition to the potential splicing in the last (Casp 1,-5,-9) blot of Figure 4C, the results presented in the corrected figure are quite profoundly different from those in the original figure. In particular, in the original figure, the Caspase 12 levels were similar in all 4 lanes, but in the corrected version, there is much less in the 4th lane. Also, in the Total Casp 1,-5,-9 (anti-FLAG) blot, the original blot showed several bands for each caspase (I guess resulting from auto-processing?), whereas the updated version instead shows remarkably clean, single bands for each protein…

  • Joshua May 31, 2013 at 12:04 pm

    Does it matter that most PEOPLE don’t express caspase-12?

    • Stewart May 31, 2013 at 1:45 pm

      Good point Joshua,

      It reminds me of those proteins that have been shown to be expressed in mice, are not essential for life, have been shown by rather odd assays to have function X (activation of a kinase, for example) in vitro and when investigated in vivo show odd effects in ONE tissue only, yet gain hefty funding for areas of medical research, such as cancer.

      Either we as a science community start to get a real grip on science-fraud or society will never reap the rewards. Afterall, the work we do is for one thing in the main – to improve human life.

  • ferniglab May 31, 2013 at 12:26 pm

    All is well, the University considers its reputation untarnished, grant funding is maintained, and the committee pats itself on the back for a job well done, etc.

    Brings to mind various combinations of words, e.g., Titanic, shifting deck chairs.

  • Junk Science May 31, 2013 at 3:17 pm

    Freeloader posted this comment on one of the previous posts about this paper and I think the picture he/she uploaded could be of interest for people who didn’t follow that particular thread.
    “In reply to michaelhbriggs (May 30, 2013 at 9:22 am):

    Indeed I noticed the sharp edge on the “Control Casp1,-5,-9″ blot too. If you adjust the contrast, you can see that this sharp line extends to the top and bottom of the blot: http://imageshack.us/photo/my-images/849/splicing.jpg/

    It certainly appears that the last band was spliced onto the blot. Quite unbelievable – I would have thought that, given the opportunity to submit a correction rather than have the paper retracted, the authors would have made sure the corrected figures were beyond reproach…”

    • michaelhbriggs May 31, 2013 at 6:43 pm

      In the Nature paper, Figure 2a, panel III looks like panel II if you rotate it clockwise 90 degrees, magnify and crop it. (Also, in the supplemental files http://www.nature.com/nature/journal/v440/n7087/suppinfo/nature04656.html panel VI of Fig 2a looks like a horizontally compressed version of panel V.)

      • pacotius June 1, 2013 at 7:44 am

        Very obviously true.
        The Supplementary Figure Legends are also extraordinary. Supp Figure 2 has 4 sections A-D, with A having seven panels i-vii.

        But the legend for Supp Figure 2 reads – and I’m not joking…

        ‘Supp. Figure S2 TNFα and IL-6 levels are not modulated by caspase-12 in response to TLR/NOD2 ligands see also Fig. 2).’

        That’s it – and there is a solitary bracket at the end. Clearly neither the authors, reviewers or editors even bothered to check that the Supplementary Figures and their legends even make sense.

        With this sort of approach to publishing, which a ten-year old could see is filled with manipulated images and incomplete information, is Nature much better than toilet reading?

  • crazyboot June 1, 2013 at 1:05 am

    There doesn’t appear to be anything suspicious about Fig 2. Note the figure legends:

    Caspase-12 expression was examined by staining tissue sections from Casp12-/- mice that either underwent CASP (panels IV–VII) or not (panels II–III) for beta-galactosidase expression…

    Also, panel VI and V are different magnifications. It is quite often for IHC of immunoflor images to be shown at different magnifications for readers.

    The authors should have had added scale bars to be clear.

    • Junk Science June 1, 2013 at 2:00 am

      I don’t agree crazyboot, since in this case they have rotated one of them 90 degrees and compressed another one as pointed out by michaelhbriggs. Bad science!

  • puzzled monkey June 1, 2013 at 12:08 pm

    I have a question. Is there a logical (licit or illicit) reason to repeatedly falsify figures when your protocol is solid and reproducible? Anybody?

    • Scrutineer June 1, 2013 at 6:17 pm

      Yes. It’s called laziness.

  • forgottenman2013 June 1, 2013 at 12:57 pm

    Reblogged this on The Firewall.

  • Post a comment

    Threaded commenting powered by interconnect/it code.