One of the things we try to do here at Retraction Watch is keep tabs on retracted work that appears again the literature. We did that twice in one day last year, once with a paper about chimps that was retracted from Biology Letters and ended up in the Journal of Human Evolution, and then again with a PLOS ONE paper on on “longevity genes” that had been retracted from Science.
Today, we have another case. In September 2010, we reported that a Cell paper, “VMA21 deficiency causes an autophagic myopathy by compromising V-ATPase activity and lysosomal acidification,” was being retracted:
Our paper reported the identification of mutations in the gene VMA21 in patients with X-linked myopathy with excessive autophagy (XMEA) and characterized the molecular mechanisms underlying the disease phenotype. Many of the figure panels in the paper summarize data from multiple experiments. We have now detected a number of errors in these panels. Although we stand by the validity of our conclusions, we believe that the most responsible course of action is to retract our paper. We are preparing an expanded version of our work for future submission. We deeply regret this circumstance and apologize to the community.
One of the original authors, Dr. Aubourg, could not be reached regarding this Retraction.
That paper has found a new home, in Acta Neuropathologica, with all of the same authors except Pauline Aubourg. Here’s the new abstract:
X-linked Myopathy with Excessive Autophagy (XMEA) is a childhood onset disease characterized by progressive vacuolation and atrophy of skeletal muscle. We show that XMEA is caused by hypomorphic alleles of the VMA21 gene, that VMA21 is the diverged human ortholog of the yeast Vma21p protein, and that like Vma21p, VMA21 is an essential assembly chaperone of the vacuolar ATPase (V-ATPase), the principal mammalian proton pump complex. Decreased VMA21 raises lysosomal pH which reduces lysosomal degradative ability and blocks autophagy. This reduces cellular free amino acids which leads to downregulation of the mTORC1 pathway, and consequent increased macroautophagy resulting in proliferation of large and ineffective autolysosomes that engulf sections of cytoplasm, merge, and vacuolate the cell. Our results uncover a novel mechanism of disease, namely macroautophagic overcompensation leading to cell vacuolation and tissue atrophy.
And here’s the retracted one, very similar:
X-linked myopathy with excessive autophagy (XMEA) is a childhood-onset disease characterized by progressive vacuolation and atrophy of skeletal muscle. We show that XMEA is caused by hypomorphic alleles of the VMA21 gene, that VMA21 is the diverged human ortholog of the yeast Vma21p protein, and that like Vma21p it is an essential assembly chaperone of the V-ATPase, the principal mammalian proton pump complex. Decreased VMA21 raises lysosomal pH, which reduces lysosomal degradative ability and blocks autophagy. This reduces cellular free amino acids, which upregulates the mTOR pathway and mTOR-dependent macroautophagy, resulting in proliferation of large and ineffective autolysosomes that engulf sections of cytoplasm, merge together, and vacuolate the cell. Our results uncover macroautophagic overcompensation leading to cell vacuolation and tissue atrophy as a mechanism of disease.
The original paper has been cited 35 times, according to Thomson Scientific’s Web of Knowledge, with 17 of those citations coming after the retraction. The new paper does not mention the retraction.
Update, 3:30 p.m. Eastern, 2/13/13: Corresponding author Berge Minassian tells us there are no significant differences between the new and retracted versions of the paper, and that they were not able to find Aubourg. We also asked whether he and his colleagues had considered mentioning in the new paper that the Cell version had been retracted.
No. That actually did not occur to me. Of course the editor of Acta and the reviewers of the paper at Acta were fully aware.
Update, 11:30 p.m. Eastern, 2/14/13: Minassian emailed to clarify that he meant that among the experiments presented again in the new paper, there were no significant differences with what was in the Cell paper. He said he did not realize were were asking about the paper as a whole, and notes that there are in fact also new experiments in the Acta Neuropathologica version.
I would read no significance (again!) into the fact that one author has not signed. In Marsailles and a long way down the paper there is no reason to believe anything other than they could not contact her.
I can’t spot anything in the old version and the new version is paywalled.
Another interesting case is this retraction:
http://www.ncbi.nlm.nih.gov/pubmed/22986374
“Retraction. Two initiation sites of early detection of colon cancer, revealed by localization of pERK1/2 in the nuclei or in aggregates at the perinuclear region of tumor cells.
[No authors listed]
Abstract
An international independent review was conducted on the above article by the editor following a complaint. The expert reviewers agreed that the interpretation of the results was not correct. The article is retracted as the results were not adequately proven. [the original article was published in the International Journal of Oncology 40: 782-788, 2012 DOI: 10.3892/ijo.2011.1268].
Retraction of
Amsterdam A, Shezen E, Raanan C, Schreiber L, Slilat Y, Fabrikant Y, Melzer E, Seger R. Int J Oncol. 2012 Mar;40(3):782-8. ”
An unusually frank rebuff I would have thought. The authors must be true believers as they seemed to have turned around and immediately republished the exact same abstract:
“Acta Histochem. 2013 Jan 25. pii: S0065-1281(12)00181-X. doi: 10.1016/j.acthis.2012.12.006. [Epub ahead of print]
Two initiation sites of early detection of colon cancer revealed by localization of pERK1/2 in the nuclei or in aggregates at the perinuclear region of the tumor cells.”
Actually not quite the same as the first version had an extra comma. But I don’t suppose the original objection concerned punctuation
Yes this was indeed interesting. Amsterdam was presented with a long list of errors in labeling figures and inaccurate statements etc, etc, to which he refused to respond appropriately and quite rightly the editor (DA Spandidos) had the article(s) – yes there was another one retracted (Use of multiple biomarkers for the localization and characterization of colon cancer stem cells by indirect immunocytochemistry. Abraham A Amsterdam, Calanit C Raanan, Letizia L Schreiber, Ora O Freyhan, Yakov Y Fabrikant, Ehud E Melzer Int J Oncol 41(5):1888 (2012) independently re-reviewed, with the result that they both were retracted by the editor. But Amsterdam evidently disagreed with this, blaming a mere ‘ specific language inaccuracy’ (see http://www.biotechniques.com/news/Independent-Review-Results-in-Cancer-Research-Retractions/biotechniques-335315.html#.UWQUlqWGZhM), although this is very far from the truth. The Acta Histochemica paper, although flawed, is however substantially changed compared to the original Int J Oncol paper.
An interesting insight into the way this group thinks, and indeed on how the Editor of Acta Histochemica (R Coleman) responds to complaints about highly inaccurate papers which are posted on line having been apparently peer-reviewed, can also be seen at http://www.sciencedirect.com/science/article/pii/S0065128113000287 and Amsterdam’s inadequate response at http://www.sciencedirect.com/science/article/pii/S006512811300029. It is quite clear that the editor of Acta Histochemica is quite unconcerned that he is publishing inaccurate papers or that he has allowed substantial changes to be made to already accepted papers when the instructions to authors clearly state ‘Kindly note that once we receive your corrections, your article is considered finalised and further amendments are no longer possible’. I wonder what the Editorial Board would think of this?
Notice that the reported effect of reduced free amino acids on mTOR is reversed in the second paper. The reduction of free amino acids downregulates mTORC1 in the first paper but upregulates the mTOR pathway (the pathway including mTORC1 and 2) in the second. So, there may be a significant difference between the two reports.
On the other hand, the real point of the work seems to be that XMEA warps the main cellular proton pump. If so, that should eventually kill muscle cells in so many different ways that the exact mechanism may be comparatively uninteresting. (I can’t see over the paywall to determine if/how the amino acid effect was actually tested, and whether there was actually a different result, in either case.)
The data is the same in both, their argument is different, and I don’t know enough about it to know the answer. In the Cell paper, they show decreased phosphorylation of p70S6 kinase. They say this means mTOR is activated, that mTOR is activated by reduced free aa, and indeed the free aa levels are reduced in their cells.
In the new paper they show decreased phosphorylation of p70S6 kinase. They say this means mTORC1 is inhibited (citing a different prior paper), that mTORC1 is inhibited by decreased free aa, and indeed free aa are reduced in their cells.
So the change relates to what pp70S6 and free aa have to do with mTOR, and they don’t interrogate mTOR directly, just report free aa and pp70S6 and draw conclusions based on different literature. Maybe the state of the field changed in 2 years, or maybe it’s a writing goof.
Hah! Thanks for confirming my dark suspicions. In either case, they’re essentially guessing.the mechanism based on a general similarity of the effect. Things are presumably better today, but, at least in my biochem days (our pipettes were graduated in sixteenths of a dram, calibrated to the Imperial Cream Cheese Standard) it was also considered particularly risky to base conclusions on protein phosphorylation levels, since there are so many wild phosphatases/kinases released by cell lysis.
http://www.ncbi.nlm.nih.gov/pubmed/?term=18824542
Fernando if you are posting papers that concern you but that have nothing to do with the topic of the main post, could you annotate your comments as “off topic”? It is confusing to me to try and search these links you post that seem to have no relationship to the main topic of the post.
Please read the author list before you write “have nothing to do with the topic of the main post.”
OK, I missed that, sorry. I didn’t realize how many authors were hidden in the list. However, this study you cite (Liu) is from Ann Arbor Michigan and it is not clear what BAM’s contribution might have been. (The studies mentioned further below are clearly from the Toronto group.)
I count 2 of the authors as being from Toronto.
It’s all for free. In the top righthand corner of the pubmed page it takes you to the original paper.
Runhua Liu1,
Lizhong Wang1,
Chong Chen1,
Yan Liu1,
Penghui Zhou1,
Yin Wang1,
Xirui Wang1,
Julie Turnbull4,
Berge A. Minassian4,
Yang Liu1,2 and
Pan Zheng1,3,*
– Author Affiliations
1Division of Immunotherapy, Department of Surgery, Program of Molecular Mechanism of Diseases and Comprehensive Cancer Center
2Division of Molecular Medicine and Genetics, Department of Internal Medicine
3Department of Pathology, University of Michigan, Ann Arbor, Michigan 48109
4Program in Genetics and Genome Biology, The Hospital for Sick Children, Toronto, Ontario, Canada
Why not add a sentence pointing out that this is a publication which shares some of the authors mentioned in this RW post?
In reply to StrongDreams February 13, 2013 at 12:47 pm
The post by little grey rabbit February 13, 2013 at 10:38 am
is not about these authors, yet you made no comment about that. Why is that?
Yes, I thought you would like this group Fernando (not the one in the original blog post BTW)
I can’t see what has caught your interest in this paper (aside from Figure 3 with phospho and unphosphorylated proteins not lining up, ie from different blots – but I have been told this is optional). But something about they way they present their blots suggests they may reward closer inspection
I did spot this one
http://endo.endojournals.org/content/144/2/612/F6.large.jpg
From 2003
Expand+Endocrinologyendo.endojournals.org
1.doi: 10.1210/en.2002-220690 Endocrinology February 1, 2003 vol. 144 no. 2 612-622
The alpha HA panel looks cobbled together.
Sorry Fernando is on topic – it is me that is off topic. Excuse my gibberish
In reply to little grey rabbit February 13, 2013 at 12:58 pm
RE: Mol Cell Biol. 2008 December; 28(23): 7236–7244.
“I can’t see what has caught your interest in this paper”.
I put it up so people might take a look. I didn’t want to influence people in what they noticed. With hindsight that was too cryptic.
Figure 5 caught my eye.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2593373/figure/f5/
I did mention to look at this figure aftter I put the paper up, but it got pushed down the page.
I put the other two papers up so people might take a look without being unduly influenced.
For the SP1 strip the eight hr of the GSK3B(+/+) and the 16 hr of the GSK3B(-/-) look the same. There may be more, but I am a bit busy.
In reply to fernando p, Fig. 5 Mol Cell Biol. 2008 December; 28(23): 7236–7244. Part A GSK3b(+/+)MEF 8hr SP1 (bottom row) looks remarkably similar to Part B GSK3b(-/-)MEF 16hr SP1.
Let figure 5 fill your computer screen.
Sit back a bit and look at both the 4a and 4b beta-actin set of panels. Try to look at both sets of panels at the same time.
Look at both 16 h panels. It may help.
Note that figure 4a and figure 4b deal with cells with different genotypes.
Take a look at figure 5.
To myself at fernando pessoa February 13, 2013 at 3:17 pm
Beta-actin set of panels in figures 5a and 5b.
To fernando pessoa himself regarding beta-actin set of panels in figures 5a and 5b (from Mol Cell Biol. 2008 December; 28(23): 7236–7244.): If you stretch the b actin in 5a vertically, (e.g. the 16hr one) it looks remarkably similar to the 16 hr one in 5b. Another round of drinks for fernando at the Brazilian Cafe!
In reply to michaelbriggs February 13, 2013 at 4:01 pm
RE: Mol Cell Biol. 2008 December; 28(23): 7236–7244
I think that all the panels in the figure 5b beta-actin set of panels are vertically stretched versions of the the corresponding figure 5a beta-actin panels (same time points). The16 h panels are the easiest to see because the bands have the grey fuzz below where they meet (most distinctive).
Please tell me what you see in the panels in Hum Mol Genet. 2005 Sep 15;14(18):2727-36.
http://www.ncbi.nlm.nih.gov/pubmed/16115820
In reply to fernando pessoa Feb 13 4:29 (below) I have looked at the figures in Hum Mol Genet. 2005 Sep 15;14(18):2727-36 and I see many splicings, low resolution, and blown out contrast especially in Figs. 2 and 5, but I have not noticed any duplications or rotations etc. After downloading the originally submitted version (ddi306v1) I have formed the opinion that the original reviewers and the editors cannot have looked at the figures very carefully, or they are in need of spectacles.
The old and new versions are very similar. On a quick scan, it appears that all of the numerical data (charts, graphs) and the EM pictures are the same, but they have replaced all the western blots and immunofluorescence pictures. A couple of the western blots in the original look slightly dodgy but it’s hard to be sure. It could also be a simple mixup with antibodies or antibodies that were not specific enough and were replaced. The figures in the new version are also larger, better labeled and easier to read, probably they had to squeeze them to get into Cell.
“were” should be “we”
Fixed, thanks.
http://www.ncbi.nlm.nih.gov/pubmed/16115820
It is by the same senior author.
http://www.ncbi.nlm.nih.gov/pubmed/21552327
It is by the same senior author
PLoS Genet. 2011 Apr;7(4):e1002037. doi: 10.1371/journal.pgen.1002037. Epub 2011 Apr 28.
PTG depletion removes Lafora bodies and rescues the fatal epilepsy of Lafora disease.
Turnbull J, DePaoli-Roach AA, Zhao X, Cortez MA, Pencea N, Tiberia E, Piliguian M, Roach PJ, Wang P, Ackerley CA, Minassian BA.
Source
Program in Genetics and Genome Biology, The Hospital for Sick Children, Toronto, Canada.
http://www.plosgenetics.org/article/info%3Adoi%2F10.1371%2Fjournal.pgen.1002037
From the text:
“Recently, DePaoli-Roach generated a mouse line deficient of PTG. In contrast to an earlier report that disruption of the PTG gene was embryonic lethal [22], the present mice are healthy and have normal lifespan [23]. Their glycogen is reduced by 30% in skeletal muscle and by 70% in brain.”.
I could find reference 22.
J Clin Invest. 2003 May;111(9):1423-32.
PTG gene deletion causes impaired glycogen synthesis and developmental insulin resistance.
Crosson SM, Khan A, Printen J, Pessin JE, Saltiel AR.
Source
Department of Medicine, Life Sciences Institute, University of Michigan School of Medicine, Ann Arbor, Michigan 48109-0650, USA.
Reference 23 did not appear in Pubmed (not the end of the world), but here:
http://professional.diabetes.org/Abstracts_Display.aspx?TYP=1&CID=55374
another person could not find the reference and was redirected:
http://www.plosgenetics.org/annotation/listThread.action;jsessionid=568F2ACDB28BCB51873BCEC2A233E09D?root=7013
I could find no evidence of the genotype (or of the lack of expression of PTG), of the the mice lacking PTG in PLoS Genet. 2011 Apr;7(4):e1002037, i.e. there is no evidence that PTG is lacking.
Reference 23 to the mice lacking PTG is an abstract.
There was no supplementary information so the evidence was not there either.
In reply to michaelbriggs February 13, 2013 at 5:56 pm
There is something not quite right. It is not just Cell. 2009 Apr 17;137(2):235-46. that had problems, but
Mol Cell Biol. 2008 December; 28(23): 7236–7244 and Hum Mol Genet. 2005 Sep 15;14(18):2727-36 also seem to have deficiencies.
At times like this I wish I could change my name to puzzledmonkey
When I looked for retractions in Hum Mol Genet I came across this one:
Hum. Mol. Genet. (2007) 16 (21): 2647.
http://hmg.oxfordjournals.org/content/16/21/2647.full
where the authors write:
“The authors deeply regret that the transgenic mice studied did not have a cDNA with a human GNE transgene containing the V572L missense mutation”
The senior author is, I believe, also an author on Cell. 2010 Sep 17;142(6):984. The retraction mentioned in this post.
I start to wonder if all is well in the state of Denmark.
Is the field unfortunate or careless?
“Is the field unfortunate or careless?”
The HMG paper was from Japan, using transgenic mice to study phenotypes associated with human allelic variations. They say that mice claimed to have one mutation (most common in Japanese population) actually have a different mutation (second-most common in Japanese population). Sounds like they were working with several gene mutations and may have made more than one line of transgenic mice and mixed up a plasmid or a breeding pair at some point. Very unfortunate and showing it is a good idea to re-test your animals from time to time.
But in the paper that is the subject of this post, Dr. Nishino in Japan is 11th out of 24 authors for a study that was done in Toronto. We don’t know what his contribution was, perhaps a cell line, or intellectual consultation about lysosomes, or something else. We know it wasn’t mice, since no mice are used in the Toronto paper. The issue with the Toronto paper is not mis-identified or mixed up plasmids or genes, but improperly manipulated images of gels.
Hard to see it as anything other than an unfortunate coincidence.
In reply to StrongDreams February 20, 2013 at 4:22 pm
Thanks for your thoughts. Those were mine too. It could all be unfortunate coincidence, but there is another paper where I Ishino is an author which is not quite right.
For example Hum Mol Genet. 2010 Jan 1;19(1):170-80.
Figure 1A. 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=2792155_ddp47701.jpg
I am having misgivings about some muscular dystrophy research in general, which does count as a field.
The reason I was looking in Hum Mol Genet is because of this paper
Hum Mol Genet. 2000 Jul 22;9(12):1843-52.
Figure 4 puzzled me.
http://hmg.oxfordjournals.org/content/9/12/1843/F4.large.jpg
Looking at this paper was prompted by the recent retraction by two of the official Optistem collegues of the first author (Yvan Torrente) those being Guilio Cossu and Emilio Clememti.
http://jcb.rupress.org/content/200/3/359.full
http://www.optistem.org/about-us/principal-investigators
A working hypothesis is that “Mother Theresa fields” may get out of hand.
I don’t see anything obvious in those two linked jpgs, but I don’t know what you were driving at.
It would not surprise me if research fields with higher sympathy/victim status had higher levels of misconduct, (possibly people trying to prove cures and give hope with weak data?)
In reply to StrongDreams February 22, 2013 at 1:11 pm
Thanks for taking a look. I appreciate your comments.
Sometimes I am a bit too cryptic for myself and had to remember why I was looking at Hum Mol Genet in the first place.
You are quite right that the 2 jpgs are not linked.
What I would ask you to do is take a look at Hum Mol Genet. 2000 Jul 22;9(12):1843-52.
Figure 4
http://hmg.oxfordjournals.org/content/9/12/1843/F4.large.jpg
which puzzled me. In particular please take a look at the beta-Actin RT-PCR and Dystrophin expression RT-PCR panels. I believe that all the samples are different therefore separate preparations. I am at loss to explain the results. I hope you can see my point.