Correction to: Oncogene (2008) 27: 145–54, published online 16 July 2007

Since the publication of our article, it has come to our attention that the legend of Fig. 2 did not acknowledge that this figure had been adapted from a published source (Irshad et al., 2004).

The legend in Fig. 2 should read as follows:

Figure 2:

Brn-3b correlates with cyclin D1 in cell lines and tumour biopsies. (A)(i)(a) Representative western blot analysis showing increased cyclin D1 protein in IMR32 cells overexpressing Brn-3b (Brn-3b+) and lower levels in cells with reduced Brn-3b (3b α-sense) compared with LTR1 (vector) control. (b) Brn-3b protein expression in stably transfected cells lines either overexpressing Brn-3b or with antisense to reduce its levels. [Adapted with permission from Figure 1 (Irshad et al., 2004)]. (ii) Quantification of cyclin D1 protein in IMR32 cells expressing different levels of Brn-3b by scanning densitometry of three independent experiments. (iii) Reduction of Brn-3b in MCF7 cells using Brn-3b antisense results in a corresponding decrease in cyclin D1 protein levels compared with vector control cells. (B)(i) Correlation of Brn-3b mRNA with cyclin D1 levels in tumour samples. qRT–PCR was used to quantify human cyclin D1 (TaqMan Gene Expression Assay, hCG2016647 (Applied Biosystems)) and GAPDH levels were used to adjust for variability. Regression analysis was carried out using Sigma plot. (i) The significant relationship between cyclin D1 and Brn-3b (R = 0.55) in NB biopsies, which was unique since (ii) shows poor correlation between Brn-3b and cyclin E mRNA in these samples (R = 0.0002). (C) Significant correlation between Brn-3b mRNA with cyclin D1 in breast cancer biopsies following qRT–PCR (R = 0.85). Breast cancer RNAs were obtained from Candis Tissue Bank (Liverpool, UK) and CR-UK Hedley Atkins Breast Pathology Laboratory (Guy’s Hospital, London). LTR, long terminal repeat; NB, neuroblastoma.