Too Much to be Nothing?
(March 24th, 2015) Already at an early age, Olivier Voinnet had achieved star status among plant biologists – until suspicions arose last year that more than 30 of his publications contained dubious images. Voinnet’s colleagues are shocked – and demand an explanation.
Several months ago, a small group of international plant scientists set themselves the task of combing through the relevant literature for evidence of potential data manipulation. They posted their discoveries on the post-publication peer review platform PubPeer. As one of these anonymous scientists (whose real name is known to Laborjournal/Lab Times) explained, all this detective work was accomplished simply by taking a good look at the published figures. Soon, the scientists stumbled on something unexpected: putative image manipulations in the papers of one of the most eminent scientists in the field, Sir David Baulcombe. Even more strikingly, all these suspicious publications (currently seven, including papers in Cell, PNAS and EMBO J) featured his former PhD student, Olivier Voinnet, as first or co-author.
Baulcombe’s research group at The Sainsbury Laboratory (TSL) in Norwich, England, has discovered nothing less than RNA interference (RNAi) in plants, the famous viral defence mechanism, which went on to revolutionise biomedical research as a whole and the technology of controlled gene silencing in particular. Olivier Voinnet himself also prominently contributed to this discovery, which certainly helped him, then only 33 years old, to land a research group leader position at the CNRS Institute for Plant Molecular Biology in Strasbourg, in his native country, France. During his time in Strasbourg, Voinnet won many prestigious prizes and awards, such as the ERC Starting Grant and the EMBO Young Investigator Award, plus the EMBO Gold Medal. Finally, at the end of 2010, the Swiss Federal Institute of Technology (ETH) in Zürich appointed the 38-year-old EMBO Member as Professor of RNA biology. Shortly afterwards, Voinnet was awarded the well-endowed Max Rössler Prize of the ETH.
Also, Lab Times reported about the young Zürich professor. There, Voinnet stated: “I hate it when people tell me what to do”. He praised the academic structures in Switzerland because the ETH would trust him completely as a scientist. Given the current suspicion of data manipulation, ETH Zürich would probably be well-advised to reconsider their blind trust for now.
The wave of public accusations against Voinnet’s publications, as reported on Retraction Watch in January 2015 and subsequently in an online Laborjournal editorial (20.01.2015), hit, with an even bigger force, also other Voinnet papers: those, which he authored after his stint in the Baulcombe lab, as last author or senior collaborator. Most publications came from his former lab at the CNRS Strasbourg but also ETH publications were soon challenged on PubPeer. Currently, a staggering number of roughly 35 Voinnet publications are under suspicion for image manipulation, including some in the elite journals Science, Nature Cell Biology, PNAS and EMBO J.
On Retraction Watch, the commenter Neuroskeptic described this as “a veritable Blotterdämmerung”, as most concerns are about seemingly duplicated blot and gel images as well as manipulative lane splicing and erased bands. Sometimes, the issues reported by the detective scientists seem to be, at first glance, hardly significant: e.g. the panels of loading controls appear to have been re-used in different contexts. Yet, without a reliable loading control, it’s impossible to conclude anything from any assay. If the loading controls have been tampered with, one has every reason to question the entire figure, if not even the key messages of the publication itself.
What is particularly noteworthy: the alleged manipulations, discovered by the detective scientists, are difficult to discern, that’s how professional they seem to have been done. One has to look closely to see the repetitive patterns or suspicious straight transition edges in the membrane background. Some blot images are also excessively bright, so that all membrane background virtually disappears, conveniently together with any possibility to scrutinise the image for integrity and possible band splicing.
Most of the anonymous PubPeer messages about inconsistencies in Voinnet’s publications are to be classified as very serious. Only a few are likely misunderstandings, created by over-eager anonymous peers: one peer actually found fault with the manufacturer-made structural pattern of the blot membrane. As for the others, it depends on your point of view. Many take them as evidence of wilful image manipulation, others protest them to be nothing more but innocent artefacts of image compression. As the plant biologist and director of the Max-Planck-Institute for Developmental Biology in Tübingen, Germany, Detlef Weigel, tweeted “Once you cut corners that badly, everything’s suspicious". Indeed, soon even Voinnet’s PhD thesis from 2001 ended up being criticised for possible irregularities on PubPeer.
Elsewhere on PubPeer, a proper quarrel arose between two commenters about a figure from the paper by Mari-Ordóñez et al. (Nat Genet, 45(9):1029-39). The whole argument could have been quickly resolved with the original uncompressed image, submitted by the authors to the journal around a year ago. Yet neither the first author, nor his boss Voinnet, nor their collaboration partner, Vincent Colot from École Normale Supérieure in Paris, chose to respond to this request of the PubPeer users and Laborjournal/Lab Times. The Nature Genetics editorial office hid behind obscure copyright restriction policies and refused to disclose the original image as well. Please note, we are talking about a published image, whose high-resolution version is for some reason to remain secret. Laborjournal/Lab Times has tried to learn more from Arturo Mari-Ordóñez’ recent doctorate thesis. Unfortunately, at the time of enquiry and several months after the official defence, he did not yet submit his dissertation. This also surprised the ETH library.
David Baulcombe, however, immediately replied to Laborjournal/Lab Times and explained to have first learned of the allegations from an anonymous e-mail shortly before Christmas 2014. He asserted that he had never encouraged his employees “to publish anything other than the results obtained without distortion or manipulation”. Former colleagues from the Baulcombe lab, Andrew Hamilton and Maria Teresa Ruiz, expressed their surprise about the suspicions raised against Voinnet. Both of them reported being “completely unaware of any misconduct on Olivier´s part” while they worked in Baulcombe´s group. However, both Hamilton and Ruiz admit that everyone worked independently on their experiments, and that the figures were generally assembled by those, who had performed the relevant experiments. Therefore, neither Hamilton nor Ruiz were able to offer more insights into Voinnet’s research attitude, beside the legendary state of messiness on his work bench.
Who was responsible?
The search among the multitude of authors for those, who were actually responsible for the criticised figures, proved rather difficult. Many contact requests remained unanswered. Still, when anyone did take a stand, the former colleagues and collaborators pointed towards Olivier Voinnet. Maria Teresa Ruiz placed the responsibility for a problematic figure in Ruiz et al. (Plant Cell, 10(6):937-46) on her co-author Voinnet. Andrew Hamilton confirmed that the criticised data in Hamilton et al. (EMBO J, 21(17):4671-9) originated from Voinnet as well. Susana Rivas admitted that the suspicious figure in Voinnet et al. (Plant J, 33(5):949-56) does contain her own data, though it was assembled by Voinnet. Another former colleague, Yvonne Pinto, declined any responsibility for the incriminated figures in the PNAS paper by Voinnet, Pinto & Baulcombe (96(24):14147-52). According to Hamilton, group leader Baulcombe was not personally involved in the compilation of figures at that time.
According to these statements, the responsibility for the allegedly manipulated images in at least four papers from the Baulcombe lab obviously was on Olivier Voinnet's side. But also for Voinnet’s publications as a group leader, Laborjournal/Lab Times could successfully narrow down the responsibilities. The head of the Donald Danforth Plant Center in Missouri, James Carrington, referred the liability for the discrepancies in two collaboratively-produced Science publications, namely Deleris et al. (313(5783):68-71) and Dunoyer et al. (328(5980):912-6) distinctly onto the Voinnet lab. Jonathan Jones, group leader at The Sainsbury Laboratory (TSL) in Norwich, made clear that the critical figure in his Science paper, Navarro et al. (312(5772):436-9), was created in Voinnet’s lab and promised to repeat the experiments independently. Former collaborators indicated that a certain likely manipulated image in Sansregret et al. (PLoS Pathogens, 9(6):e1003435) was generated in the lab of Voinnet. Edith Heard, professor at the Institut Curie in Paris, also passed the responsibility for the reported inconsistencies in two of her papers in PLoS Genetics (Ciaudo et al., 5(8):e1000620 and 9(11):e1003791) onto Voinnet and his former employee Constance Ciaudo, by now also professor at the ETH Zürich. Ciaudo is directly connected to no less than five queried Voinnet publications, also their joint paper with Heard in Cell (Chow et al., 141(6):956-69) has been flagged on PubPeer.
Several of the lambasted papers were generated entirely in the Voinnet lab. A certain publication in PNAS, authored in Strasbourg solely by Voinnet and his former PhD student Guillaume Moissiard (103(51):19593-8), is heavily plagued by accusations of data manipulation. The overwhelming majority of the directly responsible authors, either previously or currently belonging to the Voinnet lab, did not return Laborjournal/Lab Times’ requests for a statement. Some of them have since become independent group leaders or professors themselves.
Like Derrick Gibbings, presently professor in Ottawa, who, regarding the criticism of his paper in Nature Cell Biology (Gibbings et al., (9):1143-9), succinctly demanded that “this be evaluated by the journal”. The recently HFSP-funded young investigator Angélique Deleris took to PubPeer to dismiss the serious allegations of data manipulation in her Science paper and to ask that “the interpretation should not be questioned and the community should remain assured about the science”. Deleris’ name features prominently on three suspicious Voinnet publications. The same applies to her former lab colleague, Moissiard, who has since returned as a group leader to the new Zürich lab of his former boss. Peter Brodersen, now group leader in Copenhagen, is first author on two flagged publications in Science (320(5880):1185-90) and PNAS (109(5):1778-83). One name, however, appears conspicuously frequently: Patrice Dunoyer. In CNRS Strasbourg, he apparently was Voinnet’s right hand and authored many problematic publications from that lab. As Voinnet left Strasbourg for the professorship in Zürich, he handed the management of his CNRS lab over to Dunoyer. Ironically, the independent Directeur de recherché, Dunoyer, also has to face blame for a suspicious last author paper (Schott et al., EMBO J, 31(11): 2553-65). Dunoyer announced on PubPeer to be “actively working to address these issues”, together with Voinnet and the journal editors. Since then, nothing else was to be read on PubPeer from those responsible. About a certain figure in a collaborative study (Sansregret et al., PLoS Pathogens, 9(6):e1003435), which was most likely pieced together in a rather outrageous manner, Dunoyer declared: "I became aware that I used a figure that was not meant to be published". Why the said figure was created and what Dunoyer was originally planning to do with it, remains a mystery.
Laurence Drouard, head of the Strasbourg CNRS, claimed to have no authorisation to comment on this puzzling statement by Dunoyer or on any of the other, about 20, affected Voinnet publications from her Strasbourg institute. The chief of the central CNRS communications department, Brigitte Perucca, mentioned the beginnings of a CNRS investigation and promised a timely update. Yet, despite several follow-up enquiries, she never replied again. Laborjournal/Lab Times reached out to Brice Kerber, the CNRS Ombudsman responsible for such cases, who then instructed (in French!) to contact Perucca, thus sealing the circle of secrecy.
But what about Voinnet himself? Earlier, Voinnet publicly promised on Retraction Watch to investigate the alleged problems with his publications, together with his co-authors. He cancelled his participation as a keynote speaker at the SWISSPLANT 2015 Symposium in February. But at the later Keystone Symposium "RNA Silencing in Plants" in Colorado, where several prominent collaborators of his and likely also the journal editors participated, Voinnet took part. There, he purportedly “acknowledged and claimed responsibility for figure shenanigans”, as David Baltrus, a plant scientist colleague from Arizona, has tweeted. Voinnet himself declined to comment due to "important legal issues" but referred the Laborjournal/Lab Times enquiry to the ETH department of communication. The ETH Corporate Communications department in turn has, politely but very resolutely, appealed to Laborjournal/Lab Times to refrain from any further requests to the scientists or the administration, due to "internal investigations", which could not be further specified.
Laborjournal/Lab Times was, however, able to obtain more detailed information by a direct enquiry to the Department of Biology. The ETH has indeed initiated an investigative commission, according to an internal guideline (RSETHZ 415), manned by the Department Director Matthias Peter as well as two external members. The completion of the investigation is expected in April or May 2015. Till then, a gagging order is apparently in place at the ETH Zürich and elsewhere, as no-one wants to comment on the matter of Voinnet publications. In this climate, even otherwise seemingly uninvolved peers refuse to comment and express the wish to stay out of all this. As someone apparently in the know said on PubPeer: "You have to stop pestering people. Consider that once an institutional investigation has started, it is unlikely that anyone involved can comment. It can be that they can't even comment to say they can't comment."
Voinnet’s former PhD advisor David Baulcombe promised to address the problems as transparently as possible. It is important to him that “the various co-authors of the papers are not disadvantaged and that the relevant areas of science can move on”. Baulcombe has also informed his current and former employers, the University of Cambridge and The Sainsbury Laboratory (TSL) in Norwich of the accusations against his papers with Voinnet. TSL’s scientific director, Cyril Zipfel, declared on PubPeer, his institute would take the situation "extremely seriously". Furthermore, he explained that TSL “is actively discussing with all other institutions concerned, to define the best way forward to co-ordinate efforts to clarify how these irregularities occurred and to take appropriate measures”.
Baulcombe does seem to be fulfilling his promise to contact the journals. On PubPeer, he and Voinnet were able to convincingly dismantle one instance of suspected band splicing in their Cell paper (103(1):157-67) and informed the journal about their intent “to publish a clarification statement”. The chief editor of the journal The Plant Cell, Sabeeha Merchant, confirmed that “Dr. Baulcombe has indicated that he plans to submit a correction”. Should there be a suspected “ethical violation”, the journal would initiate an internal investigation, together with the American Society of Plant Biologists, which will then decide on the future of the publication from 1998 (10(6):937-46). Also, The Plant Journal and PLoS Genetics have announced their intent to investigate, according to the guidelines of the Committee on Publication Ethics (COPE).
Certain other affected journals were much less forthcoming when contacted by Laborjournal/Lab Times. On February 15th, Genes & Development (G&D) made an editorial decision by publishing a Correction to Parizotto et al. (18:2237-42), which blatantly ignored all PubPeer incriminations of band erasure and splicing but addressed a hitherto unknown image aberration instead. It is unlikely though that G&D were naively unaware. Laborjournal/Lab Times has previously informed the G&D editorial offices of the PubPeer accusations regarding this and another Voinnet publication (Azevedo et al., 24:904-15). Later on, we tried to reach out to the journal again on the matter of the recent correction. However, G&D and its chief editor did not even bother to acknowledge receiving our e-mails. Higher in the hierarchy, John Inglis, head of publishing at the Cold Spring Harbor Laboratory (CSHL), only declared that ”CSHL Press has no comment to make at this time on Dr Voinnet’s publications in Genes & Development”.
While the Voinnet issue may be over and done with as far as G&D is concerned (and maybe elsewhere), numerous collaboration partners of Voinnet are extremely worried. Everyone seems to be involved in putting together and screening the original data, in order to approach the journals. Jonathan Jones from TSL in England publicly apologised on PubPeer and promised to be “asking to see the original data from the Voinnet lab”, with the intent to publish a correction. The American James Carrington wishes to do the same. In a telephone conversation with Laborjournal/Lab Times, Carrington offered detailed insights into the full extent of the situation around the Voinnet allegations. Many of the so far reported "anomalies" are considered by Carrington as well as many of his colleagues as very serious and “unacceptable”. The "egregious" nature of some errors and the sheer number of affected publications are alarming. Carrington's lab collaborated with Voinnet on two aforementioned Science papers. The charges on PubPeer are very serious indeed: alleged data manipulation by erasure and multiple re-use of bands. Additionally: three different photographs of the same gel are apparently insidiously presented as three completely different gels. The second paper was already corrected with two Errata (328(5983):1229 and 332(6027):306), back then a previously discovered image duplication went through as an honest mistake.
Carrington asserted that scientific misconduct never took place in his lab. But he made it clear that the plant science research community took “big hits” and needs to restore its credibility, which is only possible with the necessary retrospective analysis of one’s own work. Carrington and his colleagues would now turn to re-educating employees and going through their previously published data "with a fine toothcomb". The central goal is to correct the literature but Carrington specifically places the task of assessment and decision on the nature of these corrections with the journals. Carrington sets a good example: as soon as he and his colleagues had discovered an error in a previous publication (Garcia-Ruiz et al., Plant Cell, 22(2):481-96), they contacted the journal with a correction. Moreover, Carrington has posted this information and data on PubPeer, despite the fact that this paper has been hitherto completely uncommented. This is the kind of transparency and corrective power of science one can only wish for. Carrington urges Olivier Voinnet to correct the literature appropriately. With some errors, where the correct original data can be provided, corrigenda would be sufficient. Yet, Carrington does not exclude retractions, without naming any of Voinnet publications specifically. He explains that “sometimes there are individuals who cross the red line into data manipulation”, and this could possibly have been the case here. Although Carrington leaves the final decision to the journals, he invites all authors to “do the right thing” and support the retractions in the appropriate cases. Carrington stated that he and other collaborators are currently anticipating Voinnet to approach them with the complete original raw data, the recovery and preparation of which is said to be in the final phase.
Otherwise, the community of plant biologists refrains from delivering any statements, with reference to the heated situation and on-going investigations. They first want to give Voinnet the opportunity to answer himself to the accusations. Till then, the presumption of innocence is to be maintained for him and his current and former employees. This is indeed a correct approach but unfortunately, so far, none of the criticised scientists has attempted to convincingly address the charges of data manipulation, neither on PubPeer, nor on their own institutional sites. All our emails to them have gone unanswered, so far.
Ralf Reski, professor of plant biotechnology in Freiburg, Germany, pressed in his statement to Laborjournal/Lab Times for thorough investigations into the allegations, using rigorous scientific standards. These must be done by Olivier Voinnet himself as well as by CNRS Strasbourg and ETH Zürich. The journal editorial offices would be overwhelmed with the task of a comprehensive examination, given the sheer number of flagged publications and of journals affected. According to Reski, such an enquiry should be best coordinated by EMBO, in order to ensure independence. However, EMBO Director, Maria Leptin, made clear that the ultimate responsibility to obtain the necessary documents and facts lies with the current and former employers of Voinnet and the respective journals, including EMBO Journal. EMBO, thus, wishes presently not to jump ahead of their on-going investigations. EMBO J editor-in-chief, Bernd Pulverer, stated that his house is already engaged in investigations with “due diligence” and that this “process is progressing rapidly”.
James Carrington firmly believes that science is capable of detecting the misconduct of individuals and of correcting itself. All considered, it is indeed unlikely that the Voinnet case will be quietly buried and forgotten, despite the impressive thundering silence surrounding it. Thus, the Voinnet case may indeed become a necessary purging force, at least for plant biology. The will to do so seems to be present, especially given the degree of shock and horror this incident provided to the plant science community.
Olivier Voinnet himself once said to Science Careers: “You're not there just to make papers. You're there also to bring the next generation on the way to be as good, if not much better, than you." Many of Voinnet’s first authors are now themselves senior scientists. If the allegations of data manipulation against their publications should gather further strength, it would give Voinnet’s aphorism a very cynical new meaning.
The original version of this article appeared in German in Laborjournal 3-2015. This English-language version contains updated information.
Images: ETH Zürich, Nature Cell Biology, Cell
Figure 1: One of many suspicious examples: Figure 6 c and e from Nature Cell Biology, 14:1314-21 with Olivier Voinnet as corresponding last author. Commenters on PubPeer criticise that the “TUBA” lanes of both images have been duplicated (red arrow). Also, the “Total” lanes in Figure c have clearly been vertically-mirrored and re-used as “TOTAL” in e (blue arrows).
Figure 2: Another example: Figure 6c from Cell, 95:177-87 with Olivier Voinnet as first author. PubPeer commenters suspect that the two Northern Blot lanes (green frame and magnified on the right) are the result of an illegitimate duplication.