Nature 488, 337–342 (2012); doi:10.1038/nature11331

In this Article, during the preparation of Figures 2d and 3a, we processed digital western blot scans to remove duplicate or otherwise irrelevant lanes from single-blot images. Although all excisions/mergers originated from the same gel, these figure constructions should have been explicitly pointed out. Here we present the unprocessed scans (Supplementary Information) and amended figures (Figs 1 and 2). Figure 1 of this Corrigendum shows the corrected Fig. 2d, in which a duplicate run of cell line D423-MG (lane 2 in the original) was excised between cell lines D502-MG and U343 (lanes 1 and 3 in the original) and the ensuing halves of the blot were spliced together (lanes 1 and 3 in the original blot). This is now indicated by a dashed line. Similarly, Fig. 2 of this Corrigendum shows the corrected Fig. 3a, in which for the cell line U87, an additional non-targeting short hairpin RNA control (original lanes 7 and 8) was excised with the remaining halves of the blot and merged, which is now indicated by a dashed line. We also note that in the published Fig. 3a, lanes 1 and 2 of the original U87 vinculin blot were accidentally used as the loading control for shENO2-4 (lanes 9 and 10 of the original unprocessed ENO2 blot in the Supplementary Information), and lanes 9 and 10 of the vinculin blot should have been used as the correct loading control lanes. The correct loading control lanes are now shown (Supplementary Information). None of these corrections alter the original meaning of the experiments, their results, their interpretation, nor the conclusions of the paper. We apologize for any confusion this may have caused to the readers of Nature.

Figure 1
figure 1

This is the corrected Fig. 2d of the original Article, with excision indicated.

Figure 2
figure 2

This is the corrected Fig. 3a of the original Article, with excision indicated.