Abstract
Journal statement
The authors contacted The EMBO Journal in January 2018 with a request to correct several figures. In the light of further inconsistencies uncovered by the journal and in line with journal policy, the authors’ institution was informed and asked to assess the case at the institutional level.
The journal only considers publication of a corrigendum where there is definitive evidence that the published dataset remains compelling and reliable. In cases where there are no available source data for figures with image aberrations, but where biological replicates from the same dataset are provided together with a clear understanding of how the aberrations happened, a retraction of specific subsets of the published data may be considered alongside republication of the affected figures. The corresponding author's research institution (Weizmann Institute of Science) assembled an investigative committee, which concluded that serious data manipulation is apparent in four figures of the cited paper. The committee reviewed the source data and the archived lab notebooks (authenticated versions of which were also made available to the editors of this journal, available here as Appendix Data S1) and confirmed that biological replicates from the same dataset (performed at the same time) are available that support the conclusions in the paper.
Both the investigative committee and the journal editors remain concerned by the way the figures were assembled in the original manuscript, but both parties agree that the source data and lab notebooks support the authors’ claim that the major conclusions of the paper were not affected by the manipulations. The journal appreciates the institute's open and transparent manner in handling this process, and came to its decision based on findings included in the institutional report. A summary of the report is included as Appendix Letter S1.
Given these considerations, and with agreement of the authors and the institution, the journal has agreed to retract and republish correct versions of the following panels:
Figure 2A
The published figure shows the wrong gamma‐chain gel. The gamma‐chain panel in 2A is flipped and stretched from Fig 4A pp90rsk. The authors were unable to locate the source data for the Fc(epsilon)RI gamma‐chain for the given time points (i.e. 0, 1 and 3 min) as presented in the original Fig 2A. However, source data of two additional biological replicates of the same experiment performed at the same time as the data in the published figures with an additional time point (Appendix Data S1) were found and form the basis of the replaced panel.
Figure 2A.
Figure 2B
The authors wish to correct an error in the heading of this figure. The correct heading should be: IgG − DNP (and not BSA − DNP) and G63 − DNP (and not G63 + DNP). The corresponding figure legend for Fig 2B is correct.
Figure 3A
The loading controls in the published figure are identical to those in Fig 1B of the author's 2002 paper published in the European Journal of Immunology (https://doi.org/10.1002/immu.200390011). The author's lab books (Appendix Data S1) show that the authors had performed the experiments with proper controls at the same time as the data in the published figures as displayed in the corrected figure.
Figure 3A.
Figures 4 and 5
The authors used the same Total‐Erk1/2 loading control across Figs 4A and C, and 5A and D. Figure 5A shows the correct Total‐Erk1/2 bands.
Figure 4A
This figure contains incorrectly assembled Western blot images and an incorrect loading control (Total‐Erk1/2), which is duplicated from Fig 5A. Specifically, in this experiment, the same membrane was sequentially probed with antibodies specific for pp90Rsk, pMEK and SHIP (loading control), respectively. The Western blot images corresponding to Erk‐1/2 were not found to be part of this specific experiment, but were done independently and thus should have been displayed in a separate panel. In addition, a visual divider should have been inserted into the panels labelled pErk1/2 and total Erk1/2 (right panel), so as to clearly indicate the juxtaposition of the original blots. Therefore, the authors retract Fig 4A and replace it with a corrected panel, which was prepared using the relevant source data from experiments performed at the same time as the data in the published figures present in the lab notebook (Appendix Data S1). The institution confirms that there was also a similar experiment performed at the same time as the data in the published figures paired with the correct control present in the lab book (Appendix Data S1), which was not included in the paper.
Figure 4B
The beta‐laminin panel in Fig 4B is identical to the HA‐tag panel in Fig 4C. The authors could not locate the original data for Fig 4B, but they documented data from the same experiment performed at the same time as the data in the published figures with additional time points in their lab book, confirming that these findings are consistent.
Figure 4C
The loading controls for Erk1/2 and HA‐tag in this figure are incorrect, as these were duplicated from Figs 5A and 4B, respectively. The lab books (Appendix Data S1) show similar experiments performed at the same time as the data in the published figures using biological replicates with the proper Total‐Erk1/2 loading control. These data do not show the exact experiment used in Fig 4C; however, they do show that the findings are consistent.
Figure 4A-C.
Figure 5A
This figure contains incorrectly assembled Western blot images and an incorrect loading control for PKC in its lower panel. Moreover, the Western blot images corresponding to Erk‐1/2 experiments are redundant, as they are already shown in both original and corrected Fig 4A. Therefore, we retract the Fig 5A and replace it with a corrected figure, which was prepared using the relevant source data from experiments performed at the same time as the data in the published figures present in the lab notebook (Appendix Data S1).
Figure 5A.
Figure 5C
The Western blot image shown in the lower panel of this figure contains a wrong loading control (of beta‐laminin), which was duplicated from the upper panel. In addition, the loading control in the upper panel is mislabelled (it should be Lmnb, instead of tot‐p38). Therefore, the authors wish to retract the Fig 5C and replace it with a corrected figure. The loading controls are labelled with different antibodies, and one control was flipped and stretched. The journal is correcting the labelling on the upper control panel and also correcting the lower panel; the updated correct figure was assembled based on the relevant source data from the lab notebook (Appendix Data S1).
Figure 5C.
Figure 5D
The Western blot image shown in this figure contains a wrong loading control (Erk1/2), which was duplicated from Fig 5A. The authors provided source data that show that the same membrane was probed in parallel with both p‐c‐Jun‐ and SHIP‐specific antibodies; thus, SHIP represents the correct loading control for this figure.
Figure 5D.
Author's statement
We deeply appreciate the editors’ thorough and fair handling of our case and sincerely apologise for allowing our manuscript to fall below the standard and rigorous scrutiny prior to its final submission. Specifically, we retract all the problematic figures and, wherever possible, replace them by correct figures prepared in high resolution, all based on the source data taken from the lab notebook. All authors agree with the stated corrections and figure panel retractions and affirm the integrity and authenticity of the newly published data, as outlined above.
The authors take full responsibility for their mistakes and apologise for their oversight and for any confusion that may have been caused.
Editor's statement
The above corrections were agreed upon based on detailed discussions with authors and independently with their research institution. The editors had access to authenticated facsimile versions of the lab notebooks and the full institutional report, which remains confidential. The summary of the investigation is outlined in the cover letter from the head of the institutional investigation and is included as Appendix Letter S1.
Correction to: The EMBO Journal (2006) 25: 323–334. DOI 10.1038/sj.emboj.7600932 | Published online 12 January 2006