USP26 regulates TGF‐β signaling by deubiquitinating and stabilizing SMAD7

Journal statement

The above article originally published in April 5, 2017, contains a flawed analysis in Figs 5D, 5DA and D, and EV5B–E, EV5B–EB–E. The journal is therefore retracting these panels. The authors have outlined the specific issues in their correspondence 1, published alongside this notice.

Figure 5D. Original.

 

Figure 5D. Corrected.

 

Figure EV5B–E. Original.

 

Figure EV5B–E. Corrected.

 

The journal notes the following issues:

Figure 5D, 5DA reported a statistically significant negative correlation between USP26 and phospho‐SMAD2 levels in tissue microarrays from human glioblastoma patients (r = −0.15, P < 0.0001). In response to comments from Thomas MB Ware and Hong‐Jian Zhu 2, the authors have provided a re‐analysis of their data demonstrating that no significant correlation exists between the expression of these two proteins (r = −0.15, P = 0.18).

Figures 5D, 5DD and EV5B–E, EV5B–EB–E reported Kaplan–Meier curves of glioblastoma patients, but the underlying data from the REMBRANDT database were a mixture of glioblastoma, oligodendroglioma, and astrocytoma patients.

The authors have supplied a new analysis of the REMBRANDT database using glioblastoma patients only, and the journal has decided to retract and replace these panels. Further discussion on the re‐analysis is available in the correspondence co‐published with this notice 1.

Figure 5D, 5DD. Updated legend

Kaplan–Meier curves showing that glioblastoma patients with low levels of USP26 have a significantly lower overall survival than patients with high levels of USP26; REMBRANDT, P = 0.026, n = 178. P‐value was obtained by log‐rank test. All Kaplan–Meier curves are extrapolated from datasets based on median threshold levels (high versus low).

Figure EV5B–E, EV5B–EB–E. Updated legend

Kaplan–Meier curves of glioblastoma patients with TGFBRI (B, REMBRANDT, P = 0.24, n = 178), TGFBRII (C, REMBRANDT, P = 0.336, n = 178), TGFBRIII (D, REMBRANDT, P = 0.02, n = 178), and SMAD7 (E, REMBRANDT, P = 0.039, n = 178). P‐value was obtained by log‐rank test. All Kaplan–Meier curves are extrapolated from datasets based on median threshold levels (high versus low).

The authors detail their response to Zhu et al in the correspondence 1 published alongside this notice.

Author statement

All authors agree to the retraction of original Fig 5D, 5DA and support the re‐analysis of the REMBRANDT data for original Figs 5D, 5DD and EV5B–E, EV5B–EB–E.

Partial Retraction of: EMBO Reports (2017) 18: 797‐808. DOI 10.15252/embr.201643270 | Published online 05 April 2017