Development was made aware by a reader of potential duplication of data in Fig. 2A, Fig. 5C, Fig. 6 and Fig. 7A of Development (2004) 131, 1543-1552 (doi:10.1242/dev.01050).
The journal contacted the authors who said that some of the bands in western blots were duplicated during figure compilation. After discussion with Anthi Karaiskou and Catherine Jessus, Development referred this matter to Université Pierre et Marie Curie (UPMC, now Sorbonne Université), who investigated and cleared the authors of any wrongdoing. The UPMC committee decided that the conclusions of the paper were not affected by the errors and recommended correction of the paper (full reports available at: http://www2.cnrs.fr/sites/communique/fichier/rapport_conclusions.pdf and http://www2.cnrs.fr/sites/communique/fichier/rapport_analyse_detaillee.pdf). Development's editorial policies state that: “Should an error appear in a published article that affects scientific meaning or author credibility but does not affect the overall results and conclusions of the paper, our policy is to publish a Correction…” and that a Retraction should be published when “…a published paper contain[s] one or more significant errors or inaccuracies that change the overall results and conclusions of a paper…”. Development follows the guidelines of the Committee on Publication Ethics (COPE), which state: “Retraction should usually be reserved for publications that are so seriously flawed (for whatever reason) that their findings or conclusions should not be relied upon”. The standards of figure assembly and data presentation in this paper fall short of current good scientific practice. However, given that the investigating committee at UPMC declared that the conclusions of the paper were not affected by the errors, the appropriate course of action – according to COPE guidelines – is to publish a Correction, which Development has made as detailed as possible.
Readers should note that the policy of the UPMC is that authors should retain original data for 10 years; the paper falls outside this period. The authors were unable to find all the original data, but replicates of experiments carried out at the same time showing the same results were found for most blots and several new figure panels have been assembled.
In Fig. 2A, no data for the cyclin B2 blot could be found. The authors say that as the same results are also shown in Fig. 1B, and Figs 3 and 6, the cyclin B2 blot can be removed without affecting the conclusions. The new Fig. 2A is shown below. Development has not seen the original data for these results.
Fig. 4B had unmarked splicing that the authors would like to correct. Original data could not be found for these blots but results from a replicate experiment carried out at the same time are shown.
For Fig. 5C, most of the original blots were found and the corrected figure panel is shown below. Note that original data were not found for the Cdc2 kinase activity, so this autoradiogram has been removed. The authors state that this does not affect the conclusions as Cdc2 activity is reflected by its tyrosine phosphorylation level and the graph in Fig. 5B. Lines indicating where the blots in Fig. 5A have been spliced have also been added; however, Development has not seen the original data for Fig. 5A.
In Fig. 6, original blots were found for all panels except P-MAPK and the corrected figure with lines indicating splices is shown.
In Fig. 7A, replicate results from the same experiment were found. No original data were found for the H1 kinase activity, so this has been removed. The authors state that absence of these data does not affect the conclusions because Cdc2 activity is reflected by its tyrosine 15 phosphorylation level. Lines indicating where the blots in Fig. 7B have been spliced have also been added. The new figure is shown here.
As a result of corrections to these figures, readers of Development (2004) 131, 1543-1552 (doi:10.1242/dev.01050) should ignore reference to H1 kinase activity on p. 1549. The second paragraph should now read: ‘The high H1 kinase activity generated by cyclin B1 addition in the presence of Plk1 indicates that, despite the partial phosphorylation of Cdc2, the cyclinB1-Cdc2 neocomplexes are mainly active (Fig. 5C).’ Text in the fourth paragraph should read: ‘Constitutively active Plk1 expression did not allow stage IV oocytes to respond to progesterone, as indicated by the absence of Cdc25 and Myt1 electrophoretic shift and the maintenance of Tyr15 phosphorylation of Cdc2 (Fig. 7A).’
The authors apologise to the journal and readers for these errors.
Development refers readers to other notices related to the UPMC investigation, published in our sister journal, Journal of Cell Science: