The authors of a paper suggesting that cassava, a starchy vegetable that’s a major food source in much of the developing world, could one day be turned into a food staple “capable of supplying inexpensive, plant-based proteins for food, feed and industrial applications” have retracted it, following an institutional investigation that failed to find critical supporting data.
Here’s the notice for the PLoS ONE study, “Transgenic Biofortification of the Starchy Staple Cassava (Manihot esculenta) Generates a Novel Sink for Protein:”
Following the publication of this article, the authors have been unable to confirm the presence of the zeolin gene within the transgenic cassava plants in several subsequent studies. This raises concerns about the validity of the results reported in the article.
Additionally, the Committee on Research Integrity at Donald Danforth Plant Science Center has carried out an institutional investigation which revealed that significant amounts of data and supporting documentation that were claimed to be produced by the first author could not be found. Given that the validity of the results could not be verified, and in line with the recommendation issued by the corresponding author’s institution, the authors retract the article.
The authors apologize to the readers.
The paper’s first author is Mohammad Abhary, who left the Danforth Center in mid-2011, and the corresponding author is Claude Fauquet.
The study has been cited five times, according to Thomson Scientific’s Web of Knowledge. It was funded by the Gates Foundation through its BioCassava Plus program, and cassava research is an important part of The Danforth Center’s portfolio. We know that because Ivan visited their ever-expanding greenhouse earlier this year, after being invited to give a talk on responsible conduct of research. (The Danforth Center offered to pay Ivan’s travel and an honorarium, but we declined.)
Danforth president James Carrington tells Retraction Watch that Fauquet brought the problems to his attention earlier this year, after members of Fauquet’s lab were unable to replicate the work. Questions arose shortly after the paper was published, in January 2011, when researchers tried to extend the findings. A more systematic analysis, Carrington said:
indicated that the materials that were published in the paper were not as they were described, or that the materials that were described could not be found.
By the time the issues reached Carrington’s desk, members of Fauquet’s lab had done “a fair bit of experimentation” and “additional re-analysis of the materials.” The key finding was that panel C of Figure 1 in the paper, which purported to show the presence of two proteins — p35S zeolin and pPatatin zeolin — could not be replicated. Carrington:
The transgenic plants that were supposed to contain those transgenes, plants labeled as such, did not contain what those plasmids were supposed to contain. In the re-analysis, both the plasmids and the transgenic materials were found to be lacking the key genes.
That disclosure — equivalent to an allegation of scientific misconduct, said Carrington — triggered the formation of a Research Integrity Panel, as described in Danforth’s Research Integrity Policy.
We asked Carrington what investigators thought was actually in panel C, if it was not what it seemed to be:
We could not come to conclusive proof about how the data were generated. As referenced in the retraction, however, the original data on which these figures were based, and original data on which other figures in paper were based, most of that original data could not be found. The specific route by which these were produced we could not determine.
The samples that should have contained the answer were the ones stored in the lab.
However, the materials in the tubes, and other plant material, those materials lacked what they should have had. They did not contain the zeolin genes, as indicated in the retraction. Instead they had what was essentially an empty vector. Recombinant plasmids were supposed to have a gene called zeolin, which as the paper explains was intended to increase the protein content in cassava tubers. However, when the plasmids and transgenic materials were re-analyzed during the course of in-lab investigation and the committee on research integrity investigation, indeed there were plasmids, and indeed the plants were transgenic, but they did not include the zeolin.
The results of the investigation mean that
the data do not stand, they cannot be repeated, and therefore the paper is being retracted and therefore none of the results can be validated.
However, Carrington noted, the retraction does not invalidate other work on cassava.
Indeed, it is possible to increase the nutritional content of cassava. There’s at least one other nutrient we know we can increase, a vitamin, not published yet. We can increase the resistance to viruses, all done quite reliably and reproducibly. The principle of changing the nutritional value and resistance to viruses, that principle has been established quite well…It doesn’t condemn the approach, nor some of the other work done with cassava.
The Fauquet lab has not gone back to redo the study properly, Carrington said, because the Gates grant that funded the project ended a few years ago. And the Danforth notified the Gates Foundation — which continues to fund other Center work — early in the process, sometime earlier this year, as per their research integrity policy. The disclosure — the Foundation was kept up to date throughout the investigation — has had no effect on funding from Gates, Carrington said.
Abhary, meanwhile, left the Danforth — and the U.S. — in mid-2011. He had been at the Center since early 2006, carrying out his graduate school research. Carrington notes that he did not learn of the questions about the work until Abhary had already left the Center.
The PLoS ONE paper was the only one by Abhary while he worked at the Danforth, Carrington said:
But we’re looking very carefully at secondary and tertiary instances of science that was done that might have used things that were provided. I cannot say there will never be any revelations in the future related to things during his time in the Fauquet lab.
The case is the first such investigation since the Danforth instituted its research integrity policy in 2011. It’s also the first that Toni Kutchan, who served as the center’s research integrity officer until she became director of research, has ever dealt with. About two years ago, notes Kutchan, the Danforth put additional procedures into place in response to NSF Responsible Conduct of Research requirements. That included speakers, other sessions, and
a new policy to make people more aware of requirements for the retention of data, storage of samples, so it is ultimately transparent so you can show your materials. We stepped that up more this year. We’re also discussing now, although products are not quite there, going to electronic notebook keeping.
Carrington noted:
Original documents could not be found, and proper documentation of the experiments were not found, for example, the types of things you would normally put in a notebook to explain how the experiment was done, the results and experimentation.
He added:
This really shined a bright light on the need to document research through processes that were well-known through training, and that could be tracked, whereby data were preserved for periods of time that met national standards.
The case will be the subject of a session designed to limit the likelihood of similar incidents, Kutchan said.
Carrington tells Retraction Watch:
We’ve taken this really seriously.
We have to agree, both based on the details he was willing to share, and the fact that he and Kutchan spent 45 minutes on the phone with us on short notice. If only every institution were the same.
There’s a missed opportunity for a witty post title, playing on the pronunciation of the name “Fauquet”, but I see you respect these guys, so that would probably be a low shot.
It’s too bad that they couldn’t confirm their study results– it would have been a step towards improved nutrition from cassava (and eating plant materials rather than animals is an easy way to reduce anthropogenic carbon dioxide release.) I guess that’s why the conservatives wail about global warming: we’re “telling” them they can’t eat meat…
ANYWAY, it was a highly transparent and expeditious investigation, and I’m sure you found that gratifying. The apparently responsible individual has exited, stage left. Don’t you wish Das and that oncologist guy would take a hint and skip town?
Unfortunate, and hopefully Dr Fauquet and his collaborators can redo essentially experiments and salvage something useful from this work.
But there is something problematic about this. Mohammad Abhary was a graduate student when the work relating to this paper was done. The key problem seems to be that the plasmid constructs used to make the transgenics didn’t actually contain the zeolin genes.
The implication is that the graduate student purported to have made these constructs, and seemingly convinced his supervisor and coworker that the zeolin genes were properly spliced into the plasimd. I simply can’t comprehend a scenario that the graduate student would do this study completely in isolation without showing his supervisor/coworker the results of sequencing, or results of restriction digests or some oher conclusive evidence that the zeolin genes were properly incorporated. How can that not have happened?
Presumably this stuff is part of the student’s thesis. Isn’t there some explicit and entirely standard evidence for proper incorporation of the zeolin genes there? The fact that this is apparently an example of scientific misconduct rather than error and careless record keeping would suggest not, but then how could a lack of proper characterization of plasmid constructs in a thesis pass the examiners as well as supervisor and co-workers? A little odd…
My experience is out of date, but I recall that outside grad students are sometimes hosted and funded in exchange for supplying the host lab with methods they’ve learned at their home institutions. As a result, the student functions rather independently. I don’t think Danforth is a degree-granting institution, although it’s a very fine research facility; and I assume something of the sort happened here. The downside of that arrangement, as you said, is that the grad student may not be adequately supervised. It usually works very well for all concerned (one of my kids was in that situation), but …
It is written on their website that they have recently received massive support from Gates and Monsanto foundations. As you said it is somehow hard to believe that nobody has seen anything during the evaluation of their progress. Maybe Abhary works in one of those foundations now 🙂
Because the first author also worked on transgenic cassava virus resistance projects at the Danforth ( http://www.zoominfo.com/#!search/profile/person?personId=1443778829&targetid=profile ), investigation should also be performed on other projects he has been involved in.
So presumably he must have spiked his protein preps with purified zeolin in order to get his lovely strong bands on SDS-PAGE and Western Blots?
Now that is my kind of fraud, to hell with photo-shop – this is how the true pros do it.
Of course if he @#$% cloned the zeolin gene into his constructs in the first place he wouldn’t have needed to spike anything at all. But where would have been the fun in that?
Hm. Space in the greenhouse, space in the incubator, and space in the minus 80 freezer are expensive and are in limited supply. When new projects generate materials that require space, managers look for something they can throw out. There would have been a matching set of plants and plasmids containing an empty vector (no zeolin gene) as a control to demonstrate that merely subjecting the plants to the process of transformation did not cause the increase in protein production. Maybe somebody decided to save space by tossing out unnecessary leftover materials from that research project and mistakenly tossed out the plants with the vector+zeolin and kept the plants with the empty vector. Ditto for the plasmids. This could be a labeling error or a mis-interpretation of labels, which would explain why somebody would toss the zeolin plants and the zeolin plasmids with matching labels.
The continuing investigation into subsequent projects that employed materials generated by the original project may provide some evidence about whether zeolin-transformed plants did exist at one time.
dunno about that.
It is been a long time since I did this stuff and maybe it was all only superstition anyway, but looking at his primers I was told you should put 4 or 6 bases on the 5 prime end before the restriction site or else cleavage for the sticky ends would not be so good. His restriction sites are right on the end of his primers – that alone might have made cloning difficult.
actually there may be an even more fundamental error.
Going to what I assume has the appropriate 5 and 3 prime ends of the zeolin gene
http://www.plantphysiol.org/content/136/3/3447.full#sec-12
There appears to be a mismatch in the primers used in the PLoS paper and the original
The original has:
ATGATGAGAGCAAGGG
While Danforth has:
ATGATGGAGCAAGGG
Now I disclaim high order knowledge of this gene etc, but that to me, on the first viewing, looks like it will result in a complete nonsense protein. This would leave this retraction curiously over-determined: apparently he didn’t clone anything in, but if he had it still wouldn’t worked. Perhaps a partial retraction of the retraction might be in order?
Either way it would suggest that perhaps a slightly closer level of supervision and assistance might have avoided this problem entirely resulting in a successful future career – and it can’t have been a pleasant experience for the student in question, no matter how much it may have demonstrated a character flaw.
Nonetheless, I know of plenty of labs who would have appreciated the high levels of initiative and creativity that this student showed subsequent to whatever difficulties he experienced in the initial cloning stages.
Perhaps that Harvard stem cell professor might be willing to take him on?
Doh, he was using pGEM-T vector which clones the pcr product directly in (provided you use Taq polymerase). It is very difficult to understand why someone who is properly supervised could not get something into those vectors.
The what appears to be a nucleotide deletion would still destroy any change of getting results
It just occurred to me that we’re all keying right in to the stereotype of the miscreant graduate student/lab assistant who conveniently leaves town before the fraud is discovered. The retraction notice blames the first author for missing data and documentation, although apparently not for the absence of the zeolin gene. What do you think, is it clear that the first author is really to blame? Is it necessary to look at the senior author’s other papers?
Senior author should definitely be checked since his lab has already lost some engineered trait a few years ago ( http://www.danforthcenter.org/wordpress/?page_id=395&pid=1474 ). Stories like this are definitely not good for the plant biotech sector.
Ivan/Adam: Could you please run a statistics on PLoS One retractions? I speculate that their editorial policy and editorial board are somewhat responsible for the retractions. Probably we can expect more retractions from PLoS One in the near future because of the number of papers they are handling at the moment and the acceptance rate….
I think you’d want a retraction rate per 1,000 papers published, and given how many papers PLoS ONE publishes every year — more than 13,000 in 2011 — their retraction rate is probably smaller than that of many journals, certainly for their impact factor. Neil Saunders keeps a running tab of retractions by journal here (raw numbers, not rate), and PLoS ONE isn’t even in the top 20: http://pmretract.heroku.com/journals
Wait a second… The study (Biotechnologically-Modified Cassava: Protein Absorption Relative to Casein, Li. et al., 2012) used cassava obtained from Mr. Fauquet to show that zeolin-enhanced cassava allowed mice to grow, and non-enhanced cassava resulting in weight loss and lack of growth in the control mice group. Do the findings and methodology of this study need to be checked too?!
Maybe the mice were convinced they were being fed the zeolin-enhanced cassava. Rodents are not immune to the placebo effect, it seems.
How can we be sure that these results were not fabricated also?
well they might be fabricated of course, but in the hands of a truly talented chef you can put on weight from even the most unappetising ingredients. So my guess is the results reported were genuine.
STRONG MAN:
Mashed potatoes cannot hurt you,
Mashed potatoes mean no harm
I have brought you mashed potatoes
From my mashed potato farm
MICE
Take away your mashed potatoes
Leave them in the desert to dry
Take away your mashed potatoes –
You look like shepherd’s pie
…..
FRENCHMAN:
They rejected all potaoes
For a thousand nights and days
Till a Frenchman wooed and won them
With cassave louisianaise
.
MICE
Oh my corrugated lover
So creamy and so brown
Let us fly across to St Louis
And lay our tubers down.
Art imitating Life or Life imitating Art?
that was my initial reaction also – but figure 1 looks very solid. I wonder if a simpler explanation is possible:
“Cassava roots powders (Zeolin: protein enriched cassava (MOD);
60444: wildtype cassava (CON) ) were obtained from Claude Fauquet,
Danforth center, St. Louis, MO.”
Hypothetically, someone at the Danforth centre may have been aware the Cassava was not up to scratch and taken care to fortify the powders with protein prior to sending. The paper was received in April 2012 and Abhary left in mid 2011, is it possible a 2nd fraudster is lurking behind the grassy knoll?
I recall a case like this in Germany. Some scientists reported that they could accomplish absolute asymmetric synthesis by running reactions inside a magnet of an NMR spectrometer. Following a deluge of skepticism from other groups, they repeated the experiments. They always seemed to work as long as one specific postdoc was around. It turned out he was spiking the reaction mixtures with the desired product. A similar trick was pulled off by a grad student in Sames’ group at Columbia – the perpetrator lost her degree as a result.
Conceived and designed the experiments: CMF MA NJT. Performed the experiments: MA DS GS. Analyzed the data: MA CMF DS NJT. Wrote the paper: MA NJT CMF. Generated the greenhouse and field experiments: DS GS. Produced the transgenic cassava plants: NJT
In reply to chirality September 17, 2012 at 9:30 am
Dear chirality, could you be a bit more specific about the NMR work? I think is an interesting case. We can learn from history and doing controls is always important. Often it reads like people forging things read the same playbook. When you talk about the Sames case do you mean the Sames/Sezen case?
http://blog.chembark.com/2010/12/01/sames-sezen-ori-finding/
I mention both the NMR (physics) and the Sames/Sezen case (chemistry) because these branches of science (more fundamental than biology) are under-represented for the most part.
Abhary has had to ‘eat his words’.
I like the suggestion of JudyH on Sept 15 because it would explain why the mice responded to the “fake” zeolin-spiked feed… in other words, they threw away the wrong experimental materials by accident. When they went back later to confirm it, the samples contained empty vectors because they were the controls– mislabelled as the active samples.
Or is that even possible? If littlegreyrabbit is correct, the active samples would be nonsense anyway.
There’s always the placebo effect. Or the “mashed potato farm” with mysteriously high protein levels in their “cassave louisianaise.” Thus an individual at Danforth (not Fauquet?), knowing the need for “confirmation” of the enhanced protein levels, and having access to the powdered cassava root used for the rat’s feed, added a litle zeolin… which happened to be lying around because they had to use it to supplement the feed anyway… get the picture?
Most important, the senior author saw no reason why it shouldn’t work, so it must work and there’s no reason not to tweak it a little to make sure it works. Since his instructions to the first author were oral, no record of them exists, and the gentleman has left the country anyway. And the country to which he has returned is one where he’s likely to have to “eat his words” literally as well as figuratively since the economy is so bad.
Replying to David Copperfield
Good thinking! So, NJT, that would be Nigel J. Taylor who made the transgenic cassava plants. I wonder if the Danforth Center is allowing suspicion to fall on Abhary because he conveniently is no longer there. They do specifically say that Abhary’s data and supporting documentation cannot be found, and that’s a serious problem, to be sure, but that doesn’t seem to be the underlying problem.
The main problem is that the zeolin gene is not in the plants, and apparently the investigators could not find data to support the image showing that the zeolin gene had actually been put into the cassava. That wouldn’t be Abhary’s fault.
Just wildly speculating, suppose that the empty vector was inserted in a location that disrupted expression of a gene that limited protein accumulation. Nah, that doesn’t explain why the thing worked before but doesn’t work now. Still have to go with the idea that the transformed plants were mistakenly tossed out. Although there is the possibility that the statistics were done incorrectly. Researchers can fail to notice mistakes when the results go in the direction they want to see.
Perhaps the cassava flour became infested with weevils?
That might explain the suspicious murine weight gain.
You are absolutely right. A severe weevil infestation will raise protein (and fat) levels in the infested product. You just have to bake it real hot to make sure the weevils are dead before you eat it. [I actually did this once.]
According to the publication they blotted the DNA from their “transgenic” cassava with zeolin probe. Those guys need to explain who did the SB and why it gave a positive signal. I think that retractionwatch should give them a second call to clarify those points. Just to be sure that Abhary is not a scapegoat for something much bigger.
Someone mentioned about another case coming from this lab. I have not checked but if those guys repeatedly loose traits and transgenes, they should work with another crop or simply quit science.
Although if there was a deletion in a primer leading to a nonsense sequence, a southern would come up positive, no?
Is the Mark Manary on the mouse feeding paper the same as the Mark Manary at the Danforth Center ?
http://www.danforthcenter.org/science/laboratories/Mark_Manary/
Yes seems it is the same. Everybody seems to have been evaluating those plants overZEOLously …
In the end maybe they have to close down the center if everybody has been contaminated with non-transgenic cassava :-).
I don’t understand why the paper is still up on Plos as if it is available scientific knowledge, and not taken off entirely. It doesn’t seem like a retraction.
The community continues to believe in super cassava, the retracted paper has just been quoted by a Dutch team:
Cloning and characterization of a tuberous root-specific promoter from cassava (Manihot esculenta Crantz).
Koehorst-van Putten HJ, Wolters AM, Pereira-Bertram IM, van den Berg HH, van der Krol AR, Visser RG.
Source
Wageningen UR Plant Breeding, Wageningen University and Research Center, Droevendaalsesteeg 1, 6708 PB, Wageningen, The Netherlands, [email protected].
Seems this research community is really nice and supportive.
The discovery that the transgenic casava were transformed with vector alone suggests the entirety of the Abhary paper was faked. I read the paper and was impressed – it almost seems like it would have been more work to fake it than to do the actual work (which isn’t particularly challenging). But one thing struck me – in the first figure there is a comparison of protein expression in independently isolated transgenic plants. They are all accumulating identical amounts of protein. For anyone even remotely familiar with this kind of technology, this should have been a red flag. Transgenic plants (even those with similar transgene copy number, as reportedly selected for the experiment) show a wide range of gene expression level (a normal distribution in fact). There is even a name for it (position effect). The results shown were clearly wildly improbable if not impossible.
So, the PI here and the committee members for Abhary’s PhD (at UMSL) should be feeling pretty embarrassed.