Plasticity of mammary epithelial cells in response to cyclic activation of EMT inducers

Get Complete Project Material File(s) Now! »

Cell lines and culture conditions

The immortalized non-tumorigenic human mammary epithelial MCF10a cells (ATCC, CRL-10317) were cultured in DMEM/F12 (Gibco Thermo Fischer) supplemented with 5% horse serum (Sigma), 10ng/mL cholera toxin (Sigma), 10µg/mL insulin (Sigma), 0.5µg/mL hydrocortisone (Sigma), 20ng/mL human recombinant epidermal growth factor (StemCell Technologies), 1% Penicillin/Streptomycin (Gibco), 1% L-Glutamine (Lonza), and kept in a 5% CO2 atmosphere at 37°C. Medium was renewed every two days on average.

shRNA TP53 and Brca1 knockdown

Lentiviral particles were produced by transient transfection of 293T cells using FuGENE HD (Promega) reagent and 2nd generation lentiviral system. In details, a 100mm dish of 70-80% confluent 293T cells were transfected by the vector mix (6µg psPAX2, 3µg pMD2.G and 10µg of plasmid). About 14-16h post-transfection media was replaced with 10mL basal DMEM/F12 media and the cells were grown for 48h. The virus particle containing media was collected, centrifuged at 2000g for 5 min and filtered through 0.45µm low protein binding membrane. For transduction, 60mm dishes containing 500 000 MCF10A cells were incubated overnight with 2mL of 1:1 mix of viral particle containing media and complete culture media. 2 days post transduction puromycine selection of stably transduced cells was started using a 5µg/mL concentration.

EMT/MET cycles

MCF10a cells were seeded at a density of 500 000 cells per 60mm culture dish, in complete media as described above. To induce EMT, cytokines were added to the media: TGF-β (5ng/mL) alone or in combination with TNF-α (10ng/mL) and/or IL-6 (10ng/mL). Cells were treated over a four day period. In order to induce MET, cells were cultured without cytokines for four days in between each EMT treatment phase. Media was systematically renewed every two days.

Flow cytometry

Stained cells were analysed by flow cytometry (MACSQuant ® analyzer 10, Miltenyi, Bergisch Gladbach, Germany). Data were analysed with FlowJo software (Tree Star, Ashland, OR). Freshly harvested cells were first stained with LiveDead for 20 minutes in obscurity. The cells were then either stained for EMT markers, E-cadherin and fibronectin, or for stemness markers, CD24, CD44 and CD49f. For the EMT markers, the cells were first stained with PE-conjugated anti-E-cadherin (Biolegend 324106) for 20 minutes at 4°C. The cells were then fixed using 1% PFA and permeabilised using 0.1% PBS-Triton X-100 before staining with APC-conjugated anti-fibronectin (ref??) for 20 minutes at 4°C. For the stemness markers, the cells were stained with APC-cy7a-conjugated anti-CD24 (Biolegend 311132), APC-conjugated anti-CD49f (ref??) and PE-conjugated anti-CD44 (BD Pharmingen 555479) for 20 minutes at 4°C. For analysis, events were gated to single live cells. Positivity thresholds were set using unmarked unfixed cells for stemness markers and unstained fixed cells for EMT markers.


Total RNA was extracted using the PureLink RNA kit (Invitrogen) according to the manufacturer’s instructions. cDNA synthesis was performed using High-Capacity cDNA Reverse Transcription kit (Applied Biosystems) with Oligo(dT) primers (Thermoscientific). 500ng of RNA were used for transcription. Real-time qPCR was performed with Quantstudio 5 (96 well format, Applied Biosystems) using FastStart Universal SYBR Green Master mix (Roche). Samples were used in replicates and melting curve analysis was performed for each run. Final results were expressed as relative mRNA levels calculated with the ΔΔCt method:
ΔΔCt= (2- ΔCt(test))/(2-ΔCt(control)), where
ΔCt=CT(target)-CT (housekeeping)

Immunofluorescence staining

Cells grown on 0.1% gelatine coated cover slips were fixed using a 1% paraformaldehyde at 4°C. Permeabilization was performed for 20 minutes at 4°C using a 0.1% Triton X-100 PBS solution. Non-specific antigen binding sites were blocked with a 0.1% Triton, 1% FBS PBS solution for 1 hour at room temperature. Cells were then incubated with anti-CDH1 (Santa Cruz SC21791, 1/200), anti-Oct4 (Santa Cruz SC8628, 1/400), anti-Snai1 (Cell Signaling 3879T, 1/400), anti-Zeb1 (1/400), anti-Fibronectin (ref, 1/500) and anti- α-Tubulin (ref, 1/500) antibodies overnight at 4°C. After washing, cells were incubated with secondary antibodies for 60min and the nuclei were stained with DAPI for 10 minn which was also washed with PBS later. The stained cover slips were then fixed onto glass slides using Dako fluorescence mounting medium and stored away from light at 4°C. Imaging was performed using a Nikon eclipse 90i microscope coupled with a c-HGFIE precentered fiber illuminator and a Ds-Qi1Mc camera head. The images were saved as .jpg files with an RGB color space.

Abnormal nuclei quantification

Using the fiji software, images of DAPI stained nuclei were first split into red green and blue. Only the blue channel, containing the most information, was kept for the rest of the process. The images were then smoothed using gaussian blur with a sigma value set to 4 and the background was subtracted using a rolling value of 50. Next the images were thresholded to convert them to binary black and white pictures. Any holes that appeared in the nuclei were then filled and nuclei that were fused by the thresholding were separated using the watershed function. The resulting nuclei were counted using the “analyse particles” function. The transformed images as well as numbered nuclei outlines were saved.
Due to the strong variability in abnormal nuclei shapes and sizes automated counting proved to be unreliable. Therefore, the abnormal nuclei were counted by visual appreciation, using an overlap of the original images and the numbered nuclei outlines. This enabled us to keep track of which nuclei were counted as abnormal for future reference. The proportion of abnormal nuclei is expressed as the percentage of nuclei counted by fiji that were also manually counted as abnormal.

READ  Waves propagation in fluid, solid and piezoelectric materials

BRCA1 and TP53 immunofluorescence quantification

Using the Fiji software, images of cells stained for TP53 or BRCA1 were split into red green and blue channels. The channel which contained most of the information was kept for the next steps. Images were then converted to grey scales, their background was subtracted with a rolling value of 50, and a duplicate was created. The duplicates were then converted to binary images. Holes were filled and fused nuclei were separated using the watershed function. These binary images were used as a pattern to analyse the particles of the grey scale images. The integrated density of each particle was then measured by fiji and the mean was determined. For each cell line three different fields were counted and the overall mean was calculated.


To be able to analyse the impact of BRCA1 and TP53 loss on the response of mammary epithelial cells subjected to cyclic EMT and MET induction, we first had to establish stable knockdowns for these genes in MCF10A cell lines. Two different shRNA constructs were used for both genes, using a pLKO1 backbone. Cells transfected with an empty pLKO1 backbone were used as a control. After transfection and puromycine selection of positive cells, we assessed the knockdown efficiency of each shRNA. To this end RNA was extracted for RT-qPCR quantification, and cells were seeded onto cover slips for immunofluorescence staining. Results show that the #3753 and #3754 shRNA constructs both led to a greater than 60% decrease in TP53 mRNA and protein levels (Fig. 3B, 3D, 3F). The #4984 and #4987 shRNA’s BRCA1 knockdown efficiency was more contrasted. BRCA1 mRNA levels were almost totally abolished by the #4984 shRNA. The effect of the #4987 shRNA was milder with a roughly 60% mRNA decrease (Fig. 3B, 3E, 3F). However, immunofluorescence analysis indicated a 50% decrease in BRCA1 protein levels for both the #4984 and #4987 shRNAs.
TGF-β, TGF-β + IL-6 or TGF-β + IL-6 + TNF-α cytokine treatments are potent EMT inducers To investigate the influence of cyclic activation of EMT on the mammary epithelial cell plasticity we proceeded to treat MCF10A cells with different cytokine combinations (Fig. 4). We either used TGF-β on its own (T), in combination with IL-6 (TI), or in combination with IL-6 and TNF-α (TTI). As stated in the introduction, each of these cytokines have been shown to have pro-EMT and pro-stemness effects in breast tissue. Studying their different combinations gives us a chance to identify specific consequences and potential cumulative effects on epithelial mammary cells. After each EMT induction, some of the cells were harvested and either analysed by flow cytometry or frozen for further DNA and RNA extraction and analysis. We first had to verify that the EMT induction had worked. The fluctuation of E-cadherin and fibronectin during EMT has been extensively described. E-cadherin levels drop, and fibronectin levels rise as the cells shed their epithelial properties in favour of a mesenchymal phenotype. We therefore used these two proteins as markers of EMT. There are however experimental fluctuations in E-cadherin and fibronectin levels in untreated cells (Fig. 5A). Immunofluorescence images also underline the heterogeneity of the control cells with a marked decrease in E-cadherin expression in the T2.5 cells compared to T1 (Fig. 5C). This seems to be greatly dependent on cellular confluence and E-cadherin signal can still be observed where clear cell-cell junctions are present. Furthermore, the decrease in E-cadherin observed from T1 to T3 in control cells was consistent across all cell lines indicating that a change in the culture conditions could be responsible. To take these fluctuations into consideration and simplify data interpretation, protein expression variation is represented here as fold changes relative to their respective controls at each cycle phase (Fig. 5B). Each EMT induction was successful with a strong increase observed for fibronectin and a somewhat more subdued decrease of E-cadherin. It is worth noting that for all three experimental conditions, loss of E-cadherin was very limited during the first EMT induction (Fig. 5B, EMT phase = 0.5). Immunofluorescence analysis of the cellular phenotype was performed in order to confirm the cytometry data (Fig. 5C). MCF10A pLKO1 cells are shown here, with no treatment or with TGF-β, IL-6 and TNF-α (TTI). They are representative of what we observed for the other cell lines and treatment conditions. The images show an absence of fibronectin and an abundance of E-cadherin in control cells, located primarily along the cell-cell junctions. In TTI conditions there is a strong induction of fibronectin, the cells lose their cobblestone-like appearance, spread out and assume a spindle-like morphology.

Table of contents :

Abbreviations and commercial denominations
Materials and methods
Résumé long


Related Posts