Every scientist can become a victim of a data-forging colleague or a lab member. But it is the way how they react to the discovery of research misconduct in their publications which separates truly great scientists from false heroes profiteering from a rotten system.
Another elite research institute director in Dresden is confused about the theory and practice of research integrity. Maybe this article will help.
Some scientists retract their own papers when faced with research misconduct, as the Nobel prize winner Frances Arnold very laudably did 2 years ago.
Other Nobel Prize winners, like Sir Fraser Stoddart, avoid admitting the obvious, going as far as to issuing bizarre quasi-scientific explanations to explain away what clearly looks like data manipulation. Elite journals are eager to assist them in taking the scientific community for a ride.
But then again, there are even scientists who voluntarily screen their own papers just to be on the safe side, and then retract those were data manipulations were found. Like Jonathan G Jones did in the Susana Rivas affair.
Alas, such scientists are rare. Most professors do not mind publishing things in elite journals for which a student would be expelled from the university in a blink of an eye. They all do mind getting caught though.
Will it be once again bad choices in Dresden?
The story begins. A very fake biomedical paper on the topic of endocytosis was flagged on PubPeer in April 2021.
Gaëtane Macé , Marta Miaczynska , Marino Zerial, Angel R Nebreda Phosphorylation of EEA1 by p38 MAP kinase regulates mu opioid receptor endocytosis The EMBO Journal (2005) doi: 10.1038/sj.emboj.7600799
These western blots leave no doubt about where one can stuff this paper’s main findings and conclusions.
The last author, Angel Nebreda, ICREA Research Professor and group leader at the IRB Barcelona, is a definitely not an angel of research integrity. He has a worrisome PubPeer record and his usual collaborators are often rather questionable figures, like Eric Lam, Pura Munoz-Canoves, or Manuel Benito and Angela Valverde.
As a quick detour, a small selection of other papers from Nebreda’s lab:
Anne-Claude Gavin , Aine Ni Ni Ainle , Emanuele Chierici , Margaret Jones , Angel R. Nebreda A p90(rsk) mutant constitutively interacting with MAP kinase uncouples MAP kinase from p34(cdc2)/cyclin B activation in Xenopus oocytes Molecular Biology of the Cell (1999) doi: 10.1091/mbc.10.9.2971
Ignacio Dolado , Aneta Swat , Nuria Ajenjo , Gabriella De Vita , Ana Cuadrado , Angel R. Nebreda p38alpha MAP kinase as a sensor of reactive oxygen species in tumorigenesis Cancer Cell (2007) doi: 10.1016/j.ccr.2006.12.013
Basically, not somebody one should collaborate with. Again in the EMBO Journal, note the irony that Nebreda has been EMBO member since 2003:
Ana Cuadrado, Vanesa Lafarga , Peter C F Cheung , Ignacio Dolado , Susana Llanos , Philip Cohen , Angel R Nebreda A new p38 MAP kinase-regulated transcriptional coactivator that stimulates p53-dependent apoptosis The EMBO Journal (2007) doi: 10.1038/sj.emboj.7601657
Back in June 2021, I wrote a concerned email to Nebreda’s co-author on the first paper above, Marino Zerial, Italian-born biochemist, expert in endocytosis and director at the Max Planck Institute of Molecular Cell Biology and Genetics (MPI-CBG) in Dresden, Germany. Zerial, who is also an EMBO member, and who was just completing his 2 year term as MPI-CBG scientific director, replied:
“Thank you for alerting me and my colleagues to a potential problem of violations of the principles of good scientific practice. Our Institute and the MPG in general attributes great value on following the highest standards of scientific practice. Therefore, I appreciate your efforts to identify cases of fraudulent reports. […]
Finally, you contacted me in a separate email concerning another suspicion of fraudulent activity on a manuscript in collaboration with the group of Dr. A. Nebreda. As these are not our original data, I do not have a record of these experiments. FYI: I have original data and records of my collaborators over 30 years, including my own postdoctoral work. I contacted Dr. Nebreda as well as the office of EMBO Journal to shed light into this matter. However, it will take some time, so I ask you for your patience.
I hope that this email provides some reassurance of our commitment to honour the principles of good scientific practice.“
Over half a year has passed since. I personally do not understand what exactly is unclear about the problems in Zerial’s paper with Nebreda, and why the authors simply doesn’t ask for a retraction. If Nebreda is too busy to “shed light into this matter“, will this close the case for his colleague in Dresden? Maybe the decision needs to be made elsewhere, if these two gentlemen are so undecided.
Regarding the two Nebreda papers in EMBO Journal, EMBO publishing director Bernd Pulverer told me:
There was another paper criticised on PubPeer, this time from Zerial’s own lab at MPI-CBG in Dresden. In appeared in another journal with strict image integrity policies.
Arun Pal , Fedor Severin , Barbara Lommer , Anna Shevchenko , Marino Zerial Huntingtin–HAP40 complex is a novel Rab5 effector that regulates early endosome motility and is up-regulated in Huntington’s disease The Journal of Cell Biology (2006) doi: 10.1083/jcb.200509091
Hoya Camphorifolia however thought this was not a gel duplication, but a false alarm:
“I must disagree with my learned colleague. The gist of Comment 1 is that for kDa greater than about 45, lanes 2-5 and 6-9 are the same (apart from superficial spotting and noise). To my eyes, they are different… notably, lane 9 (HAP40 / Htt4) is noticeably wider than lane 5 (Htt4).“
Maybe. But even Zerial admitted to me that this paper used to be problematic:
“In this particular case, Mike Rossner, the Executive Director of The Rockefeller University Press and Journal of Cell Biology alerted us to a potential problem of fabricated data on Fig. 1, 2 and 6, while the manuscript was in production for press. This prompted an in-depth scrutiny of the figures in question and a complete review of the original raw data, films and gels. The investigation concluded that the figures showed incorrectly collated panels but the original data and the manuscript were consistent and correct. The original films of Fig. 1, 2 and 6 as well as a letter with further explanations were sent to the JCB office. After evaluating the films, Mike Rossner concluded that “The images in the new figures clearly match the original films, and we will thus continue the production process for your manuscript using these figures.” Fig 2B was a figure of the originally submitted manuscript that we had not identified as problematic. I have checked and indeed we did scrutinize the original films that gave rise to Fig 2B and we were satisfied. Based on your observations, I will take a further in-depth look both at the data on the figures, the raw data and the conclusions of the previous investigation to determine the next steps.
As an aside but related observation from my side, the panel in question (Fig. 2B) shows the autoradiograph of in vitro–translated Htt fragments (Htt1–4) and HAP40. There is no particular mechanistic value in the Figure per se that would justify fraud (in other words, I am mystified of the motivation!).”
In December 2021, Zerial posted on PubPeer an image of the raw data (which he previously shared with me):
“Below is the scanned original autoradiograph (from 2003). This clearly shows that the bands in the published Figure are the original bands on the film, there is no splicing or duplication as was suspected.
From my point of view this confirms the results of our original investigation and the conclusion that the data published were not manipulated or fabricated.“
Maybe that Figure 2B is OK after all. But we will never find out what originally happened regarding “a potential problem of fabricated data on Fig. 1, 2 and 6“, how serious these manipulations were, and if it was the right decision to just replace the offending figures before publication and be careful not to get caught next time.
But one thing is certain: Nebreda can’t be blamed for that, in fact all authors were from the Zerial lab at the Max Planck Institute in Dresden.
Also in December 2021, Elisabeth Bik flagged another major paper from Zerial’s lab, and even tweeted it:
Takeshi Ohya , Marta Miaczynska , Ünal Coskun , Barbara Lommer , Anja Runge , David Drechsel , Yannis Kalaidzidis, Marino Zerial Reconstitution of Rab- and SNARE-dependent membrane fusion by synthetic endosomes Nature (2009) doi: 10.1038/nature08107
Zerial explained to me:
That is very confusing. How can these be the same master preparations? How can the red and blue framed Syntaxin 13 data show same master preparations when the figure legend clearly and unambiguously describes each of these bands to be derived from a different experimental setup? Also, even if the green framed data for EEA1 and Rabenosyn-5 may stem from the same preparations, why did the authors use different preparations for Rab5?
“This is actually a paper that I am very proud of (see below). To understand the figures it is necessary to go into the details of the experiments. I can provide you with the same scans of autoradiograms and explanations of the experiments I already provided to the Nature editors in 2015 (see enclosed). I have all originals files and protocols.“
“Just to guide you through:
Slide 1 shows an overview of the first author Takeshi Ohya’s folders where all protocols and original data, e.g. Western blots
Slide 2 and 3 show 2 exposures of the same blot dated 31-07-08 where the bands on Fig. 2 lane 4, 5 correspond to the band on lane 3 and the band Fig. 2 lane 6 corresponds to the band on lane 4
Slide 4 and 5 show 2 exposures of the same blot dated 09-08-08 where the bands on Fig. 2 lane 1, 3 correspond to the band on lane 5 and the band on Fig. 2 lane 2 correspond to the band on lane 4
I sent this documentation to the Nature editor (Sadaf Shadan) in 2015 who reviewed the documentation and discussed this case with her colleagues. I cite from her email “I have now had a chance to discuss the matter carefully with my colleagues. We do understand the points you make and in this case do not feel it is necessary to publish a corrigendum/addendum. Nonetheless, we would be most grateful if you’d add an online comment to the HTML version of your article on our website to briefly explain that some of the bands are the same, the reason for it and how it doesn’t affect the data/conclusions. This is to avoid future concerns being raised about Figure 2 of the paper.”
Clear as mud actually. Do the figure legends really mean something different from what they actually state? Or maybe Zerial means to say that if you use a master preparation for different experiments, the resulting western blot bands will look identical into the tiniest detail of the membrane background because the expected scientific result is somewhat similar under these two different experimental conditions?
That is of course impossible, but then again, the Max Planck Institute director did convince a Nature editor, so there.
“In response to Nature’s request, I formulated the following comment:
“Comment on Fig. 2a,b
Note that as the experiments were done on preparations of proteoliposomes that were split between different reactions, the Western blots were done on the mother mixture. In Fig. 2a, the blots of syntaxin 13 are identical in lanes 1 and 3, and lanes 4 and 5. The panels were duplicated because in lane 1-3 and 4-5 the very same proteoliposomes were used, and lane 2 originated yet from a different mixture. Syntaxin 13 is an integral membrane protein and therefore shared between different conditions, whereas Rab5, EEA1, Rabenosyn-5, etc. were added as soluble reagents and thus were blotted individually, unless the same mother reaction was split in more samples (as in panel A and B). The blots indicated by EEA1 and Rabenosyn-5 in lane 4 of Fig. 2a are duplicates of the two blots indicated by EEA1 and Rabenosyn-5 in lane 1 of Fig. 2b because the two panels show the same experiment, i.e. the sample in A was used again in B, in a different reaction (comparing without cytosol in A with cytosol in B). They were carried out at the same condition and at the same time. Importantly all proteins in the Figures, including syntaxin13 were normalized by the amount of phospholipids to maintain the ratio protein-phospholipids.”
I hope that this clarifies the matter. Please feel free to contact the Nature Editor Dr. Sadaf Shadan for confirmation.
I should also add that we have reproduced those experiments in my group over the years, by different people and with identical conclusions. Therefore, I am quite confident that the reported results in the original Nature 2009 paper are correct and reproducible. Actually, a new manuscript is in preparation with more results using a similar membrane reconstitution system as the Ohya et al., 2009.
Please let me know how I can best communicate this to Dr. Elizabeth Bik who I believe has originally raised these important issues.
First, the clarification note was never published to accompany the Nature article, so until now this information was strictly between Zerial and his Nature editor, with the rest of the scientific community left in the dark. In this regard, Professor Zerial knows perfectly well how to “communicate this to Dr. Bik“: by simply replying to her on PubPeer. Or even on Twitter! Which he hasn’t done as of yet.
And then Bik found something else.
David H. Murray , Marcus Jahnel , Janelle Lauer , Mario J. Avellaneda , Nicolas Brouilly , Alice Cezanne , Hernán Morales-Navarrete , Enrico D. Perini , Charles Ferguson , Andrei N. Lupas , Yannis Kalaidzidis , Robert G. Parton , Stephan W. Grill, Marino Zerial An endosomal tether undergoes an entropic collapse to bring vesicles together Nature (2016) doi: 10.1038/nature19326
I didn’t ask Zerial to explain these duplications, because surely those green-boxed images were the same master preparations also, or something like it, never mind what the figure legends say. Also, a triplicate experiment can just as well be a duplicate if one picture is copied and slightly shifted (red boxes). And in any case, if a Nature editor sees no reason for correction, it is not our place to ask for one.
But I did ask Zerial about this other Bik discovery, in another Nature family journal:
Stefano De Renzis, Birte Sönnichsen, Marino Zerial Divalent Rab effectors regulate the sub-compartmental organization and sorting of early endosomes Nature Cell Biology (2002) doi: 10.1038/ncb744
The first author Stefano De Renzis is now group leader at EMBL in Heidelberg, a very prestigious European research institute. I invited him and his former PI to explain that obviously photoshopped gel, where apparently a patch of background was used to cover up undesired gel bands. I also asked Zerial and his Max Planck colleagues, including the MPI-CBG Ombudsperson Elisabeth Knust, to open a research misconduct investigation, because really, how else can that figure manipulation be explained but with research misconduct?
De Renzis remained silent, but Zerial replied, on Christmas 2021:
“As I told you before, I am very much committed to the rules of good scientific practice and transparency. Therefore, also in this case, I will look into this issue. However, I need some time and help from my staff to access the archive of materials and find the original experimental data of the paper by De Renzis et al. 2002. Realistically, I will not be able to do it before the new year, because of the holidays.
I hope you can appreciate that, as I have provided you with original data and clarifications of the previous published studies in the past, ruling out scientific misconduct, it is my intention to clarify any outstanding issue that may cast doubt on the scientific integrity of my group among the scientific community. The majority of published studies were based on projects that took a period of 4-10 years, as in the Ohya et al. 2009. Results are corroborated not just by scrutiny of the results themselves but also by ensuring the data is repeated by others in my group or the observations are corroborated by alternative approaches from other labs in the meantime. This is to say that data were not obtained rapidly but only after a long and tortuous research activity involving and shared by several people in the group, and were corroborated by other groups in the field. It is not a simple collection of fabricated data to invent publishable stories. With this spirit, I am actually grateful if you help me clarify issues that may be confusing or give a wrong impression of data manipulation. The posting of information on PubPeer is a very good mechanism. Nevertheless, I cannot help but wonder to what extent you will systematically scrutinize all publications from my group. I am saying this because each investigation requires time and I am not sure that I can address all issues going back to data generated over 20 years ago.”
I explained to Professor Zerial that I am not Elisabeth Bik. And I haven’t heard from him since, and nobody else at the MPI-CBG or the Max Planck Society replied to me. I do not know if an institutional investigation was opened, or if the Max Planck Society outsources its decisions to journal editors, and otherwise allows its directors to investigate their own papers all by themselves.
As I mentioned before, there was another case of another director of another big German research institute, also in Dresden, who kept ruling out research misconduct and rejecting suggestions of an institutional investigation.
Professor Brigitte Voit resigned as her society’s Ombudsperson, her lab is now under investigation by the Leibniz Association.
The MPI-CBG director and institute’s Ombudsperson Elisabeth Knust finally replied to my emails on 14 January 2022:
“first of all, I was not aware that I had to send a reply/acknowledgement to you, since my name was put into cc in all previous e-mails, meaning just to inform me.
Second, you have sent an e-mail to me only two days ago, and you did not consider that sometimes people are busy (or gone) and do not immediately reply to e-mails [I placed the notification 3 weeks earlier, on 22 December 2021, actually. -LS] Therefore, it is not correct to write that I refused „… to admit the notification …”. or „…. that MPG and its Ombudspeople refused communication and thus also refused to admit my notification of suspected research misconduct.”
Nevertheless, I had checked the papers in the meantime, in particular the one by de Renzi et al., 2002. And I contacted Marc Stitt, the ombudsperson of the Biological Medical Section (BMS) of the MPG, who had a look into this paper as well. We both agreed that prima facie, there is something in Fig. 2c that needs to be commented on by the authors.
Marc and I will follow up this case.”
I informed Prof Knust, Prof Mark Stitt and the Max Planck Society (MPG) that I understand that aside of De Renzis et al 2002 she sees no concerns with the other papers discussed in the above article. I also announced to pass on the message to my colleague Dr Elisabeth Bik (who found those duplications) that her efforts are not appreciated by the MPG.
There was no objection, to be fair no reply at all.
PS: De Renzis’ EMBL Ombudsperson and department head Anne Ephrussi has also been notified.
I thank all my donors for supporting my journalism. You can be one of them!
Make a one-time donation:
I thank all my donors for supporting my journalism. You can be one of them!
Make a monthly donation:
Choose an amount
Or enter a custom amount
Your contribution is appreciated.
Your contribution is appreciated.DonateDonate monthly