They say that a poor workman blames his tools. If you’re a scientist and you discover your tools don’t do exactly what you thought they did, however, the right thing to do is let other scientists relying on your work know.
That’s what the University of Auckland’s Nigel Birch and colleagues did recently, after a 2012 study they published in the Journal of Neurochemistry didn’t hold up. Here’s the notice, which we’d consider a model for retractions everywhere:
The following article from Journal of Neurochemistry, ‘The serine protease inhibitor neuroserpin regulates the growth and maturation of hippocampal neurons through a non-inhibitory mechanism’ by T. W. Lee, J. M. Montgomery and N. P. Birch, published in Volume 121, Issue 4, 2012, pages 561–574 (available through www.onlinelibrary.wiley.com) has been retracted by agreement between the authors, the Journal’s Chief Editor Jörg B. Schulz, and Blackwell Publishing Ltd.
The retraction has been agreed to following the discovery of an unexpected effect of the disposable filter units on neuronal morphology. Concerns about the published data came to light following variable results in follow-up experiments investigating the mechanisms responsible for the effects reported in the article. Further investigation revealed that most of the effects attributed to neuroserpin appear to be due to a factor or factors leaching in a volume-dependent manner from disposable filter units used to sterilize the neuroserpin. Neurobasal medium that had been filtered through a 0.2 μm syringe filter resulted in increases in neurite length and reductions in neurite diameter at 2 DIV similar to those reported with neuroserpin-containing medium. The effects were seen when 2 mL of Neurobasal medium was filtered and added to the cells, but were reduced when a larger volume of medium was filtered. Medium filtration was performed to ensure the medium containing recombinant neuroserpin was sterile. As the control medium lacking neuroserpin was already sterile and it was not anticipated that medium filtration would alter neuronal growth, medium filtration was not controlled in the study. In some experiments, medium for ‘control’ wells was filtered in a larger volume, while in others the medium was not filtered at all. Therefore, an apparent effect of ‘neuroserpin’ could have been caused by presence of filter leachate in the neuroserpin conditions, which was absent in the control conditions. The authors apologize to all affected parties.
See how useful that is to anyone trying to build on this work, or use the filters in question? The journal even took care to make sure the downstream effects of the retraction were clear to readers. The paper has been cited twice, according to Thomson Scientific’s Web of Knowledge, including once by an accompanying editorial:
The retracted article was highlighted in an Editorial Highlight from Journal of Neurochemistry, ‘A role for neuroserpin in neuron morphological development’ by H.-Y. Man, and X.-M. Ma, published in Volume 121, Issue 4, 2012, pages 495–496 (available through www.onlinelibrary.wiley.com). This editorial has not been retracted, but a corrigendum published.
Dear authors and journals that insist on opaque retraction notices: Was that so hard?
This is refreshing and heartening. It is rare for real scientists to feature here and refreshing to see people actually go through their data when they have trouble reproducing a signal, rather than move on or try to cover up. Going through the minutiae of the experimental procedures and reagents to figure out where the “signal” came from IS science. This retraction enhances, rather than diminishes the reputation of the authors and the journal.
Absolutely a model for retractions and wonderful to see you include this article as a celebration of good practices, and an excellent use of your blog to share as a model of such.
Filters are often treated with surfactants, especially filters made from synthetic polymers that tend to be “non-wettable” without surfactants. And this is rarely disclosed by the manufacturer. I suspect that a lot of neuron labs will be calling this lab to ask about the specific brand, which they probably didn’t put in the retraction itself for liability reasons.
Two other examples, of many, mobile phases in columns have PEG, plasticware has plasticisers, both come out into water to different extents. We have lots of “data” on this :(. It is a real problem and wastes months. Manufacturers are usually very coy about the “additives”.
More of this will come to light as high-sensitivity untargeted metabolomics becomes accessible to the masses. Kudos (and sympathies) to the authors. This should become a case study for all first year grad students.
Continuing the “model retraction” theme, this group has followed up with a paper in the Journal of Neurochemistry detailing the contaminants and their effects. Entitled “Chemicals eluting from disposable plastic syringes and syringe filters alter neurite growth, axogenesis and the microtubule cytoskeleton in cultured hippocampal neurons” the abstract is available at http://www.ncbi.nlm.nih.gov/pubmed/25522164.