TBX3 promotes progression of pre‐invasive breast cancer cells by inducing EMT and directly up‐regulating SLUG

Abstract The acquisition of cellular invasiveness by breast epithelial cells and subsequent transition from ductal carcinoma in situ (DCIS) to invasive breast cancer is a critical step in breast cancer progression. Little is known about the molecular dynamics governing this transition. We have previously shown that overexpression of the transcriptional regulator TBX3 in DCIS‐like cells increases survival, growth, and invasiveness. To explore this mechanism further and assess direct transcriptional targets of TBX3 in a high‐resolution, isoform‐specific context, we conducted genome‐wide chromatin‐immunoprecipitation (ChIP) arrays coupled with transcriptomic analysis. We show that TBX3 regulates several epithelial–mesenchymal transition (EMT)‐related genes, including SLUG and TWIST1. Importantly, we demonstrate that TBX3 is a direct regulator of SLUG expression, and SLUG expression is required for TBX3‐induced migration and invasion. Assessing TBX3 by immunohistochemistry in early‐stage (stage 0 and stage I) breast cancers revealed high expression in low‐grade lesions. Within a second independent early‐stage non‐high‐grade cohort, we observed an association between TBX3 level in the DCIS and size of the invasive focus. Additionally, there was a positive correlation between TBX3 and SLUG, and TBX3 and TWIST1 in the invasive carcinoma. Pathway analysis revealed altered expression of several proteases and their inhibitors, consistent with the ability to degrade basement membrane in vivo. These findings strongly suggest the involvement of TBX3 in the promotion of invasiveness and progression of early‐stage pre‐invasive breast cancer to invasive carcinoma through the low‐grade molecular pathway. © 2019 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of Pathological Society of Great Britain and Ireland.

(A) Western blot analysis showing total TBX3 protein levels in 21NT stably transfected cells. 21NT cells were transfected to overexpress TBX3iso1 or TBX3iso2, or transfected with an empty vector (EV) control. Protein samples were separated by 10% SDS-PAGE and quantified by densitometry. Protein levels were normalized to Vinculin, which served as the loading control.
(B-C) Western blot analysis showing total TBX3 protein levels in 21MT-1 stably transduced cells. 21MT-1 cells were transduced with either shLUC (luciferase; off-target control) or shTBX3 which targets both TBX3 isoforms. Protein samples were separated by 10% SDS-PAGE and quantified by densitometry. Protein levels were normalized to Vinculin, which served as the loading control. (D) Total TBX3, TBX3iso1, and TBX3iso2 transcript levels were assessed by qRT-PCR in 21MT-1 transduced cells, normalized to GAPDH expression levels, and depicted as fold change relative to the shLUC control. Means derived from three biological replicates were used during analysis.
*p<0.05, **p<0.01, ***p<0.001 by one-way ANOVA with Tukey post-hoc for comparison between three groups, and student's T-test for comparison between two groups. Endothelial cells were labelled with rhodamine lectin (red), nuclei were stained with Hoechst (blue) and 21NT cells were labeled with CellTracker Green (green). Representative images of an intravascular cell (top row, white arrow) and extravasated, extravascular, cell (middle row, white arrow) are shown. Invadopodia protrusion were visualized as cell protrusions (bottom row, white arrow) extending through the endothelial layer (red). Cells were imaged in the capillary bed of the CAM using confocal microscopy.  Cells were plated on gelatin-coated coverslips, fixed, permeabilized, and stained using an anti-TKS5 antibody, followed by Alexa647-conjugated secondary antibody and Alexa488-phalloidin to stain F-actin. Single confocal slices of the ventral surface of cells are shown. Invadopodia were identified as areas of Tks5 (cyan) and actin (green) co-localization overlaying areas of degradation (black holes), as shown in the overlay.

A C + E V + T B X 3 i s o 1 + T B X 3 i s o 2
Supplementary Figure S7. Expression of EMT markers with modulation of TBX3 levels.
(A) Expression of several EMT markers was evaluated by qRT-PCR in 21NT transfectants. Expression was normalized to GAPDH expression levels, and depicted as fold change relative to the empty vector control. (B) Expression of several EMT markers was evaluated by qRT-PCR in 21MT-1 shLUC (luciferase off-target control) and shTBX3 cell lines. Expression was normalized to GAPDH expression levels, and depicted as fold change relative to the shLUC control. (C) Conditioned media was concentrated and resolved on a 10% zymogram gelatin gel. The gel was renatured and developed, and size of the proteolyzed bands was quantified by densitometry using the reverse image (as shown). Active (82 kDa The 194 genes directly bound by both TBX3 isoforms in ChIP-array data and whose transcript levels were significantly altered in expression by RNA-Seq (>1.5 fold up or down, FDR<0.05) were analyzed using the PANTHER database, conducting overrepresentation analysis and focusing on protein class. Protein classes with the lowest p-values are shown. P-values were calculated by comparing expected levels compared to input numbers in gene list.

S9) Analysis of T-box binding elements (TBEs) in TBX3-bound genes identified by ChIP-array.
Human GTCCTTGGAGGAGGTGTCAGATGGAGGAGG Bonobo GTCCTTGGAGGAGGTGTCAGATGGAGGAGG Elephant GTCCTTGGAGGAGGTGTCAGATGGAGGAGG Dog GTCCTTGGATGAGGTGTCAGATGGAGGAGG Wild boar GTCCTTGGATGAGGTGTCGGATGGAGGAGG Sea otter GTCCTTGGACGAAGTGTCAGATGGAGGAGG   (A) Cell pellets were prepared, embedded in agarose, and processed to paraffin. Sections were immunostained using anti-TBX3 antibody and counter-stained with Hematoxylin. (B-D) Subcellular fractionation was conducted in order to obtain nuclear (N) and cytoplasmic (C) fractions for 21NT transfectant, 21PT transfectant, and 21MT-1 transductant (shLUC; luciferase off-target control; shTBX3 knockdown) cell lines. Subcellular fractions were separated by 10% SDS-PAGE, and western blotting was conducted to assess TBX3 localization in nuclear (histone H4) and cytoplasmic (vinculin) fractions.