The Journal of Neuroscience‘s retraction notices often give us plenty to chew on, and a new Expression of Concern does the same.
In the notice — for a 16-year-old paper — the journal notes three cases of what certainly sounds like image manipulation, but carefully avoids calling it that:
The editors wish to express concern about the article by Ellerby HM, Martin SJ, Ellerby LM, Naiem SS, Rabizadeh S, Salvesen GS, Casiano CA, Cashman NR, Green DR, and Bredesen DE, entitled “Establishment of a Cell-free System of Neuronal Apoptosis: Comparison of Premitochondrial, Mitochondrial, and Postmitochondrial Phases” published in The Journal of Neuroscience (1997) 17:6165–6178.
The image in the top right of the panel labeled “60 min” in Figure 2A, when rotated 90° counterclockwise, appears to be more similar than would be expected to that at the top right of the panel labeled “30 min” in Figure 6A.
The image showing CPP32-stained bands in the far right lane of Figure 5A (3000 g of extract, treated with mastoparan) appears to be more similar than would be expected to that in the far right lane of Figure 8 (16,000 g of extract, treated with cytochromec/dATP).
In Figure 6C, the fodrin-stained band in the far left lane (16,000 g of extract) appears to be more similar than would be expected to that in the left lane under Bcl-2 treatment (3000 g of extract).
In response to the editors’ concerns, the corresponding author writes that the authors state “unequivocally that there was no misrepresentation of data, nor was there any scientific misconduct.”
The paper has been cited 170 times, according to Thomson Scientific’s Web of Knowledge.
The pseudonymous Clare Francis was apparently the reader who alerted the the journal to the issues, back in February. Yesterday, Peggy Mason, the chair of the ethics committee for the Society for Neuroscience, which publishes the journal, wrote Francis:
We found merit in your concerns and agreed that several images appeared to be re-used. The authors emphatically denied this. In the face of their denial, we would ordinarily have written to request an institutional inquiry. However, given the long period since publication and the re-location of authors to other institutions, we deemed it unlikely that this tact would be useful. Therefore, we published a notice of concern.
The case, Mason wrote, is “considered closed.”
Douglas R. Green, the second-to-last author of the paper, was also a co-author on two papers found to contain image manipulation by a committee at McGill. One of those papers, in Nature, was corrected, while the other was left as it is.
Would it not make sense to flag the article itself with the expression of concern, so that people newly downloading the article will see it?
Lar, You would think so, but editors don’t really care about the integrity of a scientific body of knowledge.
Its all about power, publishing, and reputation. JAMA issued its first ever “expression of concern” last year about a fraudulent hip protector study authored by Douglas Kiel of Harvard—the study text lives on with no connection to the useless expression of concern—its all a joke with these guys.
A couple of observations…
1) The paper is a “who’s who” of heavy-hitters in the early days of mitochondrial signaling/apoptosis/cytochrome c release. Dale Bredesen was until recently the head of the Buck Institute for Aging Research (Lisa Ellerby is there now). Guy Salvesen was head of the US Editorial Office for Biochem J. Doug Green has been Chair of Immunology at St. Jude for a number of years. This is certainly not a group from the backwaters of the field who faded into obscurity, but rather a group of individuals who might be colloquially referred to as superstars. Did this influence the decision in any way?
2) J. Neurosci (in my personal experience) has been somewhat reluctant to act on similar problems. In one case they agreed with me that blots were spliced together, but said that a correction “would not serve any purpose” because it didn’t change the conclusions. Also they stated that at the time of the publication the journal did not have an official policy on splicing. They do have a policy now, and claimed it would be overly punitive to retroactively apply such a policy. To me this seems rather lacking in wisdom (that’s not the way I would describe it verbally in private over a few beers, but you get my drift).
I really hate this pussyfooting phrase, “appears to be more similar than would be expected”. It’s awkward and weakens the entire expression of concern.
Yes, this is the main point. “Is the same as…” indicates misconduct. “Is similar to…” does not. Surely they can make up their mind.
I agree that an institutional investigation is inappropriate. But the paper is fundamentally unreliable and should still be retracted, any day of the week and twice on Sundays.
By the penultimate author of The Journal of Neuroscience (1997) 17:6165–6178.
Neoplasia. 2000 Nov-Dec;2(6):505-13.
Aspirin induces apoptosis through release of cytochrome c from mitochondria .
Zimmermann KC, Waterhouse NJ, Goldstein JC, Schuler M, Green DR.
Source
Division of Cellular Immunology, La Jolla Institute for Allergy and Immunology, San Diego, CA 92121, USA.
Full article can be found here: http://www.ncbi.nlm.nih.gov/pubmed/?term=11228543
Please compare the band in the 48 h HeLa lane (right-most lane) of the Cyt c panel of
figure 6B with the band in the middle lane (+ASA) of figure 5B.
I think that the treatments are different.
The lane to the left in each case is different. Which of those is correct?
Figure 5.
http://www.ncbi.nlm.nih.gov/core/lw/2.0/html/tileshop_pmc/tileshop_pmc_inline.html?title=Click%20on%20image%20to%20zoom&p=PMC3&id=1508093_neo0206_0505_fig005.jpg
Figure 6.
http://www.ncbi.nlm.nih.gov/core/lw/2.0/html/tileshop_pmc/tileshop_pmc_inline.html?title=Click%20on%20image%20to%20zoom&p=PMC3&id=1508093_neo0206_0505_fig006.jpg
Well David that didn’t take you long to convincingly show the worth of re-closing this case! I firmly agree the lid should stay tightly shut. For example your first and last authors also produced:
The role of ARK in stress-induced apoptosis in Drosophila cells. Zimmermann KC, Ricci JE, Droin NM, Green DR. J Cell Biol. 2002 Mar 18;156(6):1077-87.
PMID: 11901172
Direct link to the open access paper here:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2173462/
Figure 1B
Lower left plot “appears to be more similar than would be expected” to the lower 2nd plot despite the labelling 1.8% versus 2.2%. But they are different experiments.
Figure 3A
Microscopy image upper left “appears to be more similar than would be expected” to the lower 2nd image if we assume a partial overlap. But they are different experiments.
Figure 6C
Microscopy image upper left “appears to be more similar than would be expected” to the lower left image if we assume a partial overlap. But they are different experiments.
Microscopy image upper 3rd “appears to be more similar than would be expected” to the lower 4th image if we assume a partial overlap. But they are different experiments.
Microscopy image lower 2nd “appears to be more similar than would be expected” to the lower 3rd image if we assume a partial overlap. But they are different experiments.
However, given the long period since publication and the re-location of authors to other institutions, I deem it unlikely that an investigation would be useful. The case is considered closed.
Er, did I get that right?
Thank you for your analysis. Finding the coincidences was almost entertaining. If I can’t trust Figure 3 and Figure 6, then I can’t trust the two main messages of the paper, that inhibition of ARK resulted in “pronounced inhibition of stress-induced apoptosis” and that “cells in which cytochrome c expression was
decreased underwent apoptosis induced by stress stimuli”. The work was funded by NIH: “This work was supported by grants AI40646 and CA69381 from the National
Institutes of Health (to D.R. Green).” The paper is 13 years old, and past the six-year limitation established by NIH-ORI. However, the paper was cited last year by one of the authors: Mitochondrial pathway of apoptosis is ancestral in metazoans, 4904–4909 | PNAS | March 27, 2012 | vol. 109 | no. 13. Therefore, the exception to the six-year rule should be applied: “Subsequent use exception. The respondent continues or renews any incident of alleged research misconduct that occurred before the six-year limitation through the citation, republication or other use for the potential benefit of the respondent of the research record that is alleged to have been fabricated, falsified, or plagiarized.”
The case should not be considered closed. The coincidences are outrageous. NIH-ORI should investigate this.
Excellent work. Figure 6C especially is “more similar than would be expected” to a can of worms.
Excellent catch, someone should forward along to the editors so it can get “officially investigated” um, I mean buried.
You might enjoy certain wormy attributes in this figure.
Time has passed so that 3rd grunt on the J. Neurosci. article (that is the subject to the expression of concern discussed above) has progressed to being the senior author here.
Proteolytic cleavage of ataxin-7 by caspase-7 modulates cellular toxicity and transcriptional dysregulation.
Young JE, Gouw L, Propp S, Sopher BL, Taylor J, Lin A, Hermel E, Logvinova A, Chen SF, Chen S, Bredesen DE, Truant R, Ptacek LJ, La Spada AR, Ellerby LM.
J Biol Chem. 2007 Oct 12;282(41):30150-60.
PMID: 17646170
The worm infested figure is
http://www.jbc.org/content/282/41/30150/F2.large.jpg
The four lanes of the right panel of Fig. 2A “appear to be more similar than would be expected” to the left four lanes of Fig. 2C, though with added stars. To see this just look at those worm tracks crawling across the gel. They are a whiter shade of pale!
There is more to this figure than worms however.
In the left gel panel of Fig. 2A, lanes for caspases 9 and 10 “appear to be more similar than would be expected” to caspases 3 and 4. The splice between caspases 8 and 9 is clear, as is the reversing of the gel smile in the lower bands.
In the middle gel panel of Fig. 2A, lanes for caspases 9 and 10 “appear to be more similar than would be expected” to caspases 4 and 5. Note the vertical blemish in the middle of lanes 4 and 9.
J Biol Chem. 2003 Sep 12;278(37):34918-24. Epub 2003 Jun 24.
Kennedy’s disease. Phosphorylation of the polyglutamine-expanded form of androgen receptor regulates its cleavage by caspase-3 and enhances cell death.
LaFevre-Bernt MA, Ellerby LM
Source
Buck Institute for Aging Research, Novato, California 94945, USA.
PMID: 12824190
Figure 3.
http://www.jbc.org/content/278/37/34918/F3.large.jpg
Figure 3A. Left panel. Vertical change in background to left and horizontal change in background above band in right lane.
Figure 3D. DHT panel. Horizontal change in background above band in right lane.
Figure 1E. Vertical, straight, sharp change in signal at left end of band lane 2.
Yes Scrutineer (Nov 2 2013 3:48) in Young JE, et al. J Biol Chem. 2007 Oct 12;282(41):30150-60.
PMID: 17646170 Fig 6C bottom (Ataxin-7) panel lane 1 (pcDNA3) looks more similar than would be expected to lane 2 (Ataxin-7-10Q).
Also, (In reference to Scrutineer November 2, 2013 at 3:48) the paper J Biol Chem. 2007 Oct 12;282(41):30150-60. PMID: 17646170 has already been (mega)corrected:
“Proteolytic cleavage of ataxin-7 by caspase-7 modulates cellular toxicity and transcriptional dysregulation. VOLUME 282 (2007) PAGES 30150-30160
Jessica E. Young, Launce Gouw, Stephanie Propp, Bryce L. Sopher, Jillian Taylor,Amy Lin,Evan Hermel Anna Logvinova, Sylvia F. Chen, Shiming Chen, Dale E. Bredesen, Ray Truant, Louis J. Ptacek, Albert R. La Spada and Lisa M. Ellerby
The figures presented in the original version of this article were assembled from different gels without indicating in the figures and figure legends that such assembly had occurred as required by Journal of Biological Chemistry policy. The corrected figures are presented, and the changes are indicated below. An incorrect image of supplemental Fig. 1 was used and has been replaced on line. The conclusions of the article are unchanged by these corrections. We regret any confusion that this has caused readers.”
-funny, they mention the splicing, but not the bands and lanes that appear to be more similar than would be expected.
Thanks Michael, I had missed the earlier correction to PMID:17646170. For there to have been a correction, some noble soul must have reported issues through the “proper channels”. The case will be considered closed. Of course, as we now know, it has not yet been corrected nearly enough to change the conclusions.
That was a very good spot for the duplicated noisy lanes in Fig. 6.
May as well bring Figure 7 into the frame too. 7B, Tubulin control, the lane 2 band “appears to be more similar than would be expected” to lane 5. Lanes 4 and 6 may have a common origin too.
I wish Neuroskeptic hadn’t mentioned worms. Now I’ve got that Baha Men song going around in my head:
“Who let the worms out? Who! Who!”
After looking at the examples posted by hardman and Scrut, it does look as if serious faking has been going on.
But what can the journal do? Without a finding of misconduct and with the stern denial by the authors, the editor will be hard pressed to justify a retraction.
I think the original institution(s) has the responsibility to investigate. Current employers will be correct to state that they don’t have the ability to investigate previous misconduct at a different institution. Also, there might still be some petty vindictive scoundrels at the old department willing to agitate for an investigation.
Surely the thing would be for Clare Francis to take the ethics chairs letter to the administration of the original institution herself. Their name is on the paper(s) and ultimately it is their responsibility.
The correct way to deal with this (problems in papers spread across a number of years and institutions) is to package it all up and send a big file to the ORI. Then they can figure out how to go about investigating – which institutions, who is ultimately responsible, etc With ORI oversight, what the journals or institutions want to do is supposed to be a moot point.
The only downside is their 6 year statute-of-limitations. The way I interpret this is they will only bring action for federally funded research published within the past 6 years. But, that doesn’t mean they have to ignore “evidence” from earlier times when building a case against something within the SOL time-frame. So, if you’re sending a big pack of evidence to the ORI, so long as at least one of the papers in there is within that 6 year limit, they’re obliged to look at the whole picture.
OK, well, get on with it…
There are exceptions to the six-year limitation:
Sec. 93.105 Time limitations.
(a) Six-year limitation. This part applies only to research misconduct occurring within six years of the date HHS or an institution receives an allegation of research misconduct.
(b) Exceptions to the six-year limitation. Paragraph (a) of this section does not apply in the following instances:
(1) Subsequent use exception. The respondent continues or renews any incident of alleged research misconduct that occurred before the six-year limitation through the citation, republication or other use for the potential benefit of the respondent of the research record that is alleged to have been fabricated, falsified, or plagiarized.
(2) Health or safety of the public exception. If ORI or the institution, following consultation with ORI, determines that the alleged misconduct, if it occurred, would possibly have a substantial
adverse effect on the health or safety of the public.
(3) “Grandfather” exception. If HHS or an institution received the allegation of research misconduct before the effective date of this part.
I am confused by this one.
J Biol Chem. 2010 May 28;285(22):16632-42. doi: 10.1074/jbc.M109.095083. Epub 2010 Mar 22.
Inducible dimerization and inducible cleavage reveal a requirement for both processes in caspase-8 activation. Oberst A, Pop C, Tremblay AG, Blais V, Denault JB, Salvesen GS, Green DR.
Source
Department of Immunology, St. Jude Children’s Research Hospital, Memphis, Tennessee 38105, USA.
PMID: 20308068
http://www.jbc.org/content/285/22/16632.full.pdf+html
Please compare bands lanes 2, 3 and 4 Anti-hsp 90 panel figure Supplementary fig. 3
with bands lanes 1,2 and 3, respectively, at top Casp-8 panel figure 3.
Supplementary figures.
http://www.jbc.org/content/suppl/2010/03/22/M109.095083.DC1/jbc.M109.095083-1.pdf
I am also confused by this one.
J Biol Chem. 2001 Sep 7;276(36):33869-74. Epub 2001 Jul 11.
Coupling endoplasmic reticulum stress to the cell death program. Mechanism of caspase activation.
Rao RV, Hermel E, Castro-Obregon S, del Rio G, Ellerby LM, Ellerby HM, Bredesen DE
Source
Buck Institute for Age Research, 8001 Redwood Blvd, Novato, California 94945, USA.
PMID: 11448953
Figure 5.
http://www.jbc.org/content/276/36/33869/F5.large.jpg
Figure 5a. There is a vertical, straight grey streak/vertical change in signal between the upper 1/3rd of the lanes 2 and 3, but not between the lower 2/3rds of lanes 2 and 3.
Correction for J Biol Chem. 2010 May 28;285(22):16632-42 published 2014
http://www.jbc.org/content/289/10/6838.full
10.1074/jbc.A109.095083
Some things which struck me as problematic.
J Biol Chem. 2004 May 7;279(19):20211-20. Epub 2004 Feb 23.
Inhibition of calpain cleavage of huntingtin reduces toxicity: accumulation of calpain/caspase fragments in the nucleus.
Gafni J, Hermel E, Young JE, Wellington CL, Hayden MR, Ellerby LM.
Source
The Buck Institute for Age Research, 8001 Redwood Boulevard, Novato, California 94945, USA.
PMID: 14981075
http://www.jbc.org/content/early/2004/02/23/jbc.M401267200
Figure 2A. Vertical changes in background between lanes 1,2 and 3. There are only 3 lanes.
Figure 5A. Vertical changes in background between left and right lanes in 2nd and 6th panels, but not in 3rd,4th and 5th panels.
Figure 6B. PARP panel. Vertical, straight change in signal at right end band lane 2. Near vertical, change in signal at left end band lane 4.
Any reason you chose to spare Fig. 4? The blots there have so many joins they look like a patchwork quilt.
In reply to Scrutineer November 3, 2013 at 4:01 am
Thanks. I got a bit tired of the sudoku puzzle.
“Some things which struck me as problematic”. There were others.
People say that splicing was more common in the last century, but usually for reasons.
Proc Natl Acad Sci U S A. 1998 Apr 28;95(9):4997-5002.
Bax directly induces release of cytochrome c from isolated mitochondria.
Jürgensmeier JM, Xie Z, Deveraux Q, Ellerby L, Bredesen D, Reed JC.
Source
Program on Apoptosis and Cell Death Research, The Burnham Institute, 10901 North Torrey Pines Road, La Jolla, CA 92037, USA.
Figure 1.
http://www.ncbi.nlm.nih.gov/core/lw/2.0/html/tileshop_pmc/tileshop_pmc_inline.html?title=Click%20on%20image%20to%20zoom&p=PMC3&id=20202_pq0983396001.jpg
Figure 1A. Left panel (lanes 1 to 4). Vertical change in background between lanes 1 and 2. There are only 4 lanes.
Middle panel (lanes 5 and 6). Top band lane 5 has vertical, straight right edge. There are only 2 lanes in the panel.
SUP and PEL panels (lanes 7 and 8). Vertical changes background between left and right lanes. There are only 2 lanes in each panel.
Figure 1C. Right panel. Vertical change in background between lanes. There are only 2 lanes in the panel.
Erratum appears.
Chem Biol. 2010 Nov 24;17(11):1189-200. doi: 10.1016/j.chembiol.2010.08.014.
Identification and evaluation of small molecule pan-caspase inhibitors in Huntington’s disease models.
Leyva MJ, Degiacomo F, Kaltenbach LS, Holcomb J, Zhang N, Gafni J, Park H, Lo DC, Salvesen GS, Ellerby LM, Ellman JA.
Source
Department of Chemistry, University of California, Berkeley, CA 94720, USA.
PMID: 21095569
Erratum in Chem Biol. 2013 May 23;20(5):742.
http://www.cell.com/chemistry-biology/fulltext/S1074-5521%2813%2900176-2
Original figure 5.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3035168/figure/F5/
Proc Natl Acad Sci U S A. 1999 May 11;96(10):5752-7.
Release of caspase-9 from mitochondria during neuronal apoptosis and cerebral ischemia.
Krajewski S, Krajewska M, Ellerby LM, Welsh K, Xie Z, Deveraux QL, Salvesen GS, Bredesen DE, Rosenthal RE, Fiskum G, Reed JC.
Source
The Burnham Institute, La Jolla, CA 92037, USA.
PMID: 10318956
Figure 4.
http://www.ncbi.nlm.nih.gov/core/lw/2.0/html/tileshop_pmc/tileshop_pmc_inline.html?title=Click%20on%20image%20to%20zoom&p=PMC3&id=21932_pq1090339004.jpg
Figure 4D. Bcl-XL panel. Band left lane has vertical right edge.
Casp9 panel. Vertical, darker streaks at right edges bands left panel.
David, I really wish we had someone with such a clinical eye in plant science to look at the gels in plant science-related papers. I have tried to look at some of the figures you point out as critically as possible, but as a layman, I most likely would never have seen some of these errors, which might explain why the editors or “peers” may not have picked up on these errors either, during peer review. Unfortunately, in plant science, few are willing to retrospectively look at these issues, even less those with talent and skills, as they are all too busy doing research, or fulfilling other functions. Are you aware of any useful guidelines that could help non-specialist researchers detect problems with gel images? I will soon be making a wide call for PPPR among plant science peers, so I am trying to prepare some guidelines about how post-publication peer review could be achieved more easily.
Paper by penultimate author coming under renewed scrutiny: https://pubpeer.com/publications/11901172
https://pubpeer.com/publications/11901172
Figure 1B.
http://imgur.com/we9FUQI