Low HER2 expression in normal breast epithelium enables dedifferentiation and malignant transformation via chromatin opening

Abstract

Overexpression of the HER2 protein in breast cancer patients is a predictor of poor prognosis and resistance to therapies. We used an inducible breast cancer transformation system that allows investigation of early molecular changes. HER2 overexpression to similar levels as those observed in a subtype of HER2-positive breast cancer patients induced transformation of MCF10A cells and resulted in gross morphological changes, increased anchorage-independent growth of cells, and altered the transcriptional programme of genes associated with oncogenic transformation. Global phosphoproteomic analysis during HER2 induction predominantly detected an increase in protein phosphorylation. Intriguingly, this correlated with chromatin opening, as measured by ATAC-seq on acini isolated from 3D cell culture. HER2 overexpression resulted in opening of many distal regulatory regions and promoted reprogramming-associated heterogeneity. We found that a subset of cells acquired a dedifferentiated breast stem-like phenotype, making them likely candidates for malignant transformation. Our data show that this population of cells, which counterintuitively enriches for relatively low HER2 protein abundance and increased chromatin accessibility, possesses transformational drive, resulting in increased anchorage-independent growth in vitro compared to cells not displaying a stem-like phenotype.

Keywords: In vitro; Breast; Cancer; Chromatin; Epigenetics; Stem.

Fig. 1.

HER2 protein overexpression is sufficient to induce in vitro transformation. (A) Schematic of multi-omic analysis and soft functional assays performed with their respective timelines as MCF10A cells undergo in vitro transformation. ATAC-seq, assay of transposase-accessible chromatin using sequencing; scRNA-seq, single-cell RNA sequencing. (B) HER2 protein expression analysis by western blotting in MCF10A cells infected with inducible HER2 lentiviral particles and cultured in various concentration of doxycycline for 24 h. GAPDH was used as a loading control. n=2. (C) MCF10A^HER2 and control cells were cultured in 3D over 9 days. Control cells formed spherical acini, which increased in size over time. MCF10A^HER2 cells formed flat projecting cells of complex masses, typical of transformed cells. Images captured by a confocal, LSM 510 microscope. Scale bars: 50 µm. n=3. (D) Cell migration and invasion was analysed through the 8 µm pores of transwell membranes over a 16 h period of chemotactic migration towards full serum medium. The ability for cell invasion was measured in collagen or Matrigel-coated transwells. Migration ability was measured in using uncoated wells. Statistical significance was calculated using unpaired two-tailed Student's t-test. *P<0.05, **P<0.01; ns, not significant. n=3. (E) Colony growth of MCF10A^HER2 and control cells in 0.3% ultra-pure agarose over 3 weeks. Five different-size colonies from ImageJ analysis were quantified. Representative microscopic images of colonies stained with Crystal Violet after 3 weeks are shown on the right. Statistical significance was calculated using unpaired two-tailed Student's t-test. *P<0.05, **P<0.01, ***P<0.001. Images are at 1.6× magnification. Scale bars: 1000 μm. n=3.

Fig. 2.

HER2 promotes in vitro transformation through increase in signalling and widespread chromatin opening. (A) Volcano plots depicting the phosphoproteome upon HER2 protein expression at 0.5, 4 and 7 h compared to control cells. Statistical significance is shown as log2 fold change for HER2>1.5, P<0.05, and log2 fold change for GFP<5, P>0.05. The plot shows the phosphosites that are significantly changing upon HER2 protein induction but not significantly changing in the GFP cells at the same time. Those with the highest increase or decrease in fold change are labelled. n=3. (B) Bar graph depicting the number of detected phosphosites increasing or decreasing in phosphorylation in the phosphoproteomic analysis at the indicated time points analysed. Statistical significance is shown as log2 fold change for HER2>1.5, P<0.05, and log2 fold change for GFP<5, P>0.05). (C) Differential accessibility (log2 fold change>0.5, FDR-corrected P<0.05) between MCF10A^HER2 and control cells, plotted against the mean reads per region. Cells were grown in 3D cell culture from 0 to 48 h, and ATAC-seq was performed on their acini. Heatmap shows chromatin accessibility across all time points for each replicate in cells expressing HER2 or GFP (controls). n=3. (D) Fraction of total regions that are differentially accessible (up peaks) or inaccessible (down peaks) in early or late type comparisons. ‘Early’ time point represents data from 0, 1, 4 and 7 h combined. ‘Late’ time point represents data from 24 h and 48 h combined. Log2fold>2, FDR-corrected P<0.05. (E) Gene Ontology (GO) categories for biological processes for differential peaks that are significantly up (log2fold change>0.5, FDR-corrected P<0.05) for the early MCF10A^HER2/early MCF10A^CTRL cells.

Fig. 3.

Abnormal HER2 expression shows overlapping genes/transcription factors identified in multiomics data. (A) Distance to closest transcriptional start sites (TSSs) of all differentially accessible regions in the early and late cell types. The bars represent only those regions that are upstream of the TSS. ‘Open sea’ refers to regions that are at least 50 kb or more upstream of the TSS. (B) Enrichment of transcription factor recognition sequences in differential ATAC-seq peaks comparing MCF10A^HER2 and control cells based on HOMER analysis. Down peaks=log2fold<−2, FDR-corrected P<0.05. HOMER analysis using the accessible (up) peaks can be found in Fig. S2C. (C) Venn diagram showing the number of differentially accessible regions that are shared between the up (open) and down (closed) peaks in the early and late samples. Up peaks=log2fold>2, FDR-corrected P<0.05. Down peaks=log2fold<−2, FDR-corrected P<0.05. (D) scRNA-seq was performed in 2D cell culture on MCF10A cells with HER2 induction from 0 to 72 h (3 days). Heatmap summarises some of the most highly and lowly expressed genes with the induction of HER2 gene. (E) Insertion tracks of samples at example regions. This signal is an average signal of three replicates of combined time points into either ‘early’ samples or ‘late’ samples. Differentially open regions are marked with arrows.

Fig. 4.

Low HER2 expression leads to increased transformation, stemness and chromatin accessibility. (A) Proposed simplified breast epithelial hierarchy present in human mammary glands. (B) Cells were analysed by flow cytometry, and HER2-positive cells were separated into three subpopulations of low, medium and high HER2 overexpression, as indicated. The enrichment of stem markers is shown as a proportion of the total number of cells exhibiting MUC1-negative and EPCAM-negative phenotype. The proportion of cells shown here shows the overlap between MUC1-negative and EPCAM-negative cells, all of which were subsequently 100% CD24 negative. n=3. (C) MCF10A^HER2 cells were flow sorted into the labelled subtypes, and HER2 expression analysis in MCF10A cells by western blotting was performed. GAPDH was used as a loading control. The bottom 20% of HER2-expressing cells were labelled as ‘HER2 low’ cells (blue); the top 20% of HER2 expressing cells were labelled as ‘HER2 high’ cells (red). The middle population (35%) was labelled as ‘HER2 med’ (orange). HER2-negative cells are highlighted in green based on HER2-negative control cells. N=3. (D) HER2 expression was induced for 3 days, and cells were sorted based on HER2 expression into low, medium and high HER2 expression. 5000 cells from each condition were plated into ultra-pure agarose to investigate their in vitro transformative potential. Results are plotted as box plots from three biological replicates. Unpaired two-tailed Student’s t-test was performed to compare ‘HER2 med’ and ‘HER2 high’ groups to the ‘HER2 low’ group; P-values are displayed on the graph. One-way ANOVA was performed to determine statistical significance. The boxes represent interquartile range, and the whiskers indicate the minimum and maximum. n=3. (E) MCF10A^HER2 cells were sorted into the three subtypes. ATAC-seq libraries were prepared and sequenced. DiffBind was used to analyse the differentially accessible regions, plotted as percentage of open or closed regions. n=3. (F) Heatmap shows genes of interest that are consistently differentially expressed in at least three of the four time points analysed upon HER2 overexpression. Blue rectangles represent genes that are downregulated; red rectangles represent genes that are upregulated. The white rectangles show lack of differential expression for that specific time point. Genes are only listed if the statistical significance had FDR-corrected P<0.05. Importance of genes highlighted in red is mentioned in the text.