Aberrant expression of LMO4 induces centrosome amplification and mitotic spindle abnormalities in breast cancer cells

Abstract

The LIM-only protein, LMO4, is a transcriptional modulator overexpressed in breast cancer. It is oncogenic in murine mammary epithelium and is required for G2/M progression of ErbB2-dependent cells as well as growth and invasion of other breast cancer cell types. However, the mechanisms underlying the oncogenic activity of LMO4 remain unclear. Herein, we show that LMO4 is expressed in all breast cancer subtypes examined and its expression level correlates with the degree of proliferation of such tumours. In addition, we have determined that LMO4 silencing induces G2/M arrest in cells from various breast cancer subtypes, suggesting that LMO4 action in the cell cycle is not restricted to a single breast cancer subtype. This arrest was accompanied by increased cell death, amplification of centrosomes, and formation of abnormal mitotic spindles. Consistent with its ability to positively and negatively regulate the formation of active transcription complexes, overexpression of LMO4 also resulted in an increase in centrosome number. Centrosome amplification has been shown to prolong the G2/M phase of the cell cycle and induce apoptosis; thus, we conclude that supernumerary centrosomes mediate the G2/M arrest and cell death in LMO4-deficient cells. Furthermore, the correlation of centrosome amplification with genomic instability suggests that the impact of dysregulated LMO4 on the centrosome cycle may promote LMO4-induced tumour formation.

Figure 1

LMO4 mRNA is highly expressed in BRCA1 mutant and ER-negative breast cancers. Expression of LMO4 mRNA was assessed in cohorts of breast cancers using publicly available gene expression profiles. Graphs depict the mean LMO4 mRNA levels ± standard deviation. (A) Comparison of LMO4 mRNA levels in two independent studies showed a 1.5-fold and 2.4-fold increase in LMO4 in basal-like cancers compared to non-basal-like tumors in the Richardson (n=38) (29) and Farmer (n=43) (28) studies, respectively. (B) LMO4 mRNA was also evaluated in sporadic breast cancers (n=97) and tumors associated with BRCA1 mutations (n=18). When compared to sporadic cancers, LMO4 mRNA was increased 3.9-fold in tumors with BRCA1 mutations (30). (C) LMO4 expression was assessed in ER-positive (ER+) and ER-negative (ER−) breast cancers in several independent studies: van de Vijver (n=295), GSE2109 IGC/expO (n=232), Minn (n=99), Wang (n=286), Richardson (n=39) and Chin (n=118) (,–34). LMO4 was consistently upregulated 1.4- to 1.6-fold in ER- tumors when compared to ER+ cancers. * p<0.05.

Figure 2

LMO4 expression is elevated in tumors with high mitotic indices. (A, left panel) A cohort of triple negative (n=18), ErbB2-positive (ER−/ErbB2+ n=14) and ER-positive (ER+/ErbB2− n=16) breast tumors were evaluated for LMO4 protein expression by immunohistochemistry. Two tumors of each subtype are shown with arrows identifying LMO4 positive cells. In triple negative tumors, nearly all tumor cells were positive for LMO4. Scale bar = 100 μm. (right panel) The intensity (on a scale of 0–3, with 3 being the highest) and extent (percent of LMO4 positive tumor cells) of LMO4 nuclear stain was assessed. The final tumor score was the product of intensity and percentage of positive cells, resulting in a scale that ranged from 0 to 30. Bars are the mean histopathological score for each set of tumors and error bars depict the group’s standard deviation. * p<0.05. (B, left panel) Tumors of each subtype (ER+/ErbB2 n=4; ErbB2+/ER− n=4 and triple negative n=5) were stained for LMO4 (green) and the proliferation marker Ki67 (red). (right panel) Scatter plot analysis of the percent of tumor cells with Ki67 and intense LMO4 staining. Subtypes are represented by different symbols. Correlation coefficient, R²=0.76.

Figure 3

LMO4 regulates G2/M progression in luminal and basal-like breast cancer cells. (left panels) LMO4 expression was transiently silenced in two luminal-like (MCF-7, T47D, Panels A and B, respectively) and two basal-like (MDA-MB-468, BT-549; Panels C and D, respectively) cell lines using a previously validated siRNA (13). A representative western blot is shown for each cell line demonstrating LMO4 silencing in cells transfected with LMO4-specific siRNA (LMO4-KD) compared to cells transfected with a non-silencing siRNA control (NS). β-actin was used as loading control. (middle panels) Seventy-two hours after siRNA transfection with control (NS) or LMO4-directed siRNA (LMO4-KD), cells were harvested and stained with propidium iodide to analyze DNA content by FACS. The proportion of cells at each of the phases of the cell cycle (subG1, G1, S and G2/M) is depicted in the bar graph ± standard deviation. (right panels) Cells were transfected with siRNA and harvested every 24 hrs thereafter for 6 days. The growth of LMO4 knock-down cells was then compared to control cells (NS) by directly counting the number of viable cells at each time. * p<0.05 compared to NS transfected cells.

Figure 4

Alterations in LMO4 levels induce centrosome amplification. (A–D) LMO4 was silenced in the luminal-like MCF7 (A) and T47D (B) as well as basal-like MDA-MB-468 (C) breast cancer cells via an LMO4 targeted siRNA. LMO4 was also overexpressed in T47D cells using retroviral infection. (D) After manipulation of LMO4 levels, cells were immunostained with an antibody to the centrosomal protein γ-tubulin. The number of centrosomes/cell was then counted and compared between cells with altered LMO4 levels [either loss of LMO4 (LMO4-KD), A–C, or LMO4 overexpression (LMO4-OE), D], and control cells [cells transfected with either non-silencing (NS) siRNA or empty expression vector (vector)]. Representative images from a total of 3 independent experiments/cell line are provided for each cell type with arrows pointing to centrosomes. Normal cells have either 1 or 2 centrosomes/cell. Bars represent the relative centrosome amplification ± standard deviation compared to the control cells in three independent experiments.

Figure 5

Altered LMO4 levels induce abnormal mitotic spindles. (A–C) MCF-7 (A) and T47D (B) cells were transfected with either control siRNA (NS) or LMO4-specific siRNA (LMO4-KD), and synchronized using aphidicolin. T47D cells overexpressing LMO4 (LMO4-OE) or an empty vector control (vector) were also synchronized (C). After removal of aphidicolin, cells were immunostained with an antibody to α-tubulin and counterstained with DAPI for visualization of the mitotic spindle and DNA, respectively. The spindles were blindly scored for their symmetry and organization. Equal segregation of DNA was considered and percent of cells with abnormal spindles was normalized to that of control (either NS or empty vector transfected) cells. The average relative number of abnormal mitotic figures is shown with error bars depicting standard deviations. Three independent experiments were performed for each cell line.

Figure 6

LMO4 localizes to the centrosome. MDA-MB-468 cells were synchronized with aphidicolin and immunostained for both LMO4 and the centrosome protein γ-tubulin. DAPI was use to stain DNA. Cells undergoing mitosis as well as cells in interphase were analyzed for co-localization of LMO4 and γ-tubulin. Images of individual stains are shown in black and white, with overlapping staining shown in yellow.