Retraction of 19-year-old Nature paper reveals hidden cameras, lab break-in, evidence tampering
We’ve often found that when some authors refuse to sign retraction notices, there’s a much bigger story than terse notices let on. And a retraction in this week’s Nature of a 19-year-old paper is a shining example of that.
Here’s the brief notice for “Oligosaccharide ligands for NKR-P1 protein activate NK cells and cytotoxicity,” a 1994 paper by researchers from the UK and the Czech Republic that had already been subject to a 1996 correction:
We wish to retract this Article owing to an inability to reproduce the results. This retraction has not been signed by K.B. and A.F., and M.P. is deceased (J.O’B. cannot be traced).
K.B. is Karel Bezouska, then working in the lab of corresponding author Ten Feizi at The Glycosciences Laboratory at Northwick Park Hospital, Middlesex, UK. Bezouska, it turns out, was found by an ethics committee at Charles University (Google translation of a January 2013 press release from Czech) to have:
…most likely committed the Scientific Misconduct or “dangerous and irresponsible deviations from accepted practice how to conduct research.” This Scientific Misconduct occurred repeatedly. It’s a shame, because prof. Bezouška was considered a brilliant biochemist, one of the founders of proteomics in the Czech Republic, an excellent speaker and a devoted worker. Publication activity prof. Bezouška was high – more than 100 articles in peer-reviewed international journals. No doubt there was contained a significant amount of high-quality experimental work, unfortunately in some cases contained with high probability and non-recurrent and unprintable data.
Perhaps most astonishing is that Bezouska broke into a lab refrigerator so he could tamper with samples being used to try to replicate the experiments during the investigation:
One of the students carried out verification tests, which revealed that the experimental material stored in the fridge was handled without the knowledge of laboratory personnel. After this finding was in the room once installed camera system to track events before the refrigerator. Cameras revealed that on the night of the 19th to 20.3. twice broke into the room and 21.3 do it again broke prof. Bezouška and manipulated the experimental material in the fridge. The ethics committee after hearing all witnesses, including hearing prof. Bezouška, concluded that prof. Bezouška manipulated the samples with the highest likelihood of recurrence of artificially prepared the positive results that should confirm the binding of carbohydrates to protein NKR-P1. Positive results reported prof. Bezouška regarding custody of carbohydrates to protein NKR-P1 has repeatedly failed to replicate or reproduce more independent laboratories.
For Feizi, now at Imperial College, London, it has been a long 19 years trying to correct the scientific record. A 2011 email from her to other scientists — available on the website of Česká Pozice, a Czech newspaper that has been covering the case extensively — tells the story:
a) Re: His experiments in my lab on rat NKR-P1 published in JBC2 and Nature3 in 1994, which could not be reproduced, during the course of 4 years, with fresh proteins Karel sent us, nor with a de-novo construct5’6 generated in my lab: Karel indicated that it had been decided that he would be giving details of his protocols, to you, Vladimir, for possible attempts at repeating them.
It went on:
My response to a) was that it remained very desirable for Karel to determine why, during his time in my lab, he could repeatedly make preparations of recombinant rat NKRP-1 and get the ligand-binding data published2,3, whereas my colleagues could not reproduce them either with the very tiny mounts of the protein that he sent us after his departure, nor with recombinant protein de-novo generated in my lab during the subsequent 4 years5,6. Some of the putative ligands are commercially available and could be used in any new attempts. Complete retraction should be considered of the ligand binding data in Nature paper beyond the correction that I submitted and is on record in PubMed3.
The end of Feizi’s email suggests she has been trying hard to correct the scientific record, but that Bezouska has been a barrier:
To be honest, I doubt I would be able to interest members of my group, past and present, several of whom devoted 4 years, to what amounted to unproductive research trying to reproduce the contents of the JBC2, and Nature papers3. There were also data on human CD69 protein that were published in BBRC1 and other dramatic cell signalling data on this protein that were almost accepted for publication in Cell, but I withdrew the paper as soon as major issues of reproducibility were encountered by members of my lab. Apart from attempts in my lab, I spent several working days in Prague with Karel and an experienced post doc, Christine Galustian, working with coded samples. All results with CD69 were negative in contrast to remarkable results Karel reported and was prepared for us to submit to Cell.
I should mention that following formal investigation in 1999 by a Committee involving the Chief Executives of the Medical Research Council and of the Wellcome trust at which they interviewed Karel, who attended with support from his supervisor, the late Miloslav Pospíšil, it was suggested by the Committee that these issues of reproducibility should be resolved scientifically.
My group indeed made attempts over 4 years. In the end, all we could do was to have a retraction paper in JBC5, a partial retraction termed ‘correction’ in Nature3, and a retraction paper BBRC4 retracting the published observations1 on CD69. Karel and Miloslav declined at that time to retract the Nature paper as a whole, this is why, by rule, I could only ‘correct/retract’ the parts that my group could evaluate in the absence ofbioactive recombinant NKRP-1.
Where Feizi refers to “retraction paper,” she means follow-up studies (we’ve linked to them; one is in a different journal than the original) showing the lack of reproducibility. Bezouska’s name is not on either of those retraction papers.
Neither is his name on a 1999 correction to the 1994 JBC paper to which Feizi refers in her email. That correction begins:
As we have been unable to reproduce (1) the effects of oligosaccharides on killing by natural killer (NK) cells of the rat (2), the need has arisen to re-examine the carbohydrate-binding properties reported for recombinant soluble forms of NKR-P1A, which is a disulfide-linked homodimeric transmembrane protein of lectin type at the surface of NK cells and NK-like T cells, and is an activator of cytotoxicity (3, 4). Attempts to generate the bacterially expressed soluble forms of the monomeric carbohydrate-recognition domain (CRD)1 (5) and the dimeric full-length extracellular part (2, 5) of the rat NKR-P1A have proven unsuccessful2 in contrast to previously reported data (5).
The correction then goes on to describe other experiments.
The Nature paper has been cited 255 times, according to Thomson Scientific’s Web of Knowledge.
Update, 12:45 Eastern, 7/8/13: Feizi sent us a detailed comment:
As you will have gathered a Correction was published previously, and at that time I wanted to fully retract the Article, but the Czech co-authors did not agree as they were convinced of the validity of their data. Nature’s policy at the time was to require that all authors agree to a formal retraction.
Members of my group and various independent scientists have never been able to reproduce the results. About a year ago I was made aware of a case of scientific misconduct against Professor Bezouska and decided to approach Nature again to reconsider the retraction. We decided that I should wait until all investigations were complete.
The case of scientific misconduct has been proven and Prof Bezouska has been dismissed from his post. A subsequent court case for defamation has just come to an end.
This case has been covered in the Czech press. I attach two English translations and an official statement from the Institute of Microbiology in Prague.