Abstract
Background/Aim: Breast cancer, one of the most prevalent cancers globally, is marked by its cellular heterogeneity. A key aspect of breast cancer research is understanding the distinct morphological features of cancerous and non-cancerous cells, which could serve as potential targets for novel therapeutic interventions. In this light, our study aimed to comprehensively analyze the morphological features of the MCF10A and MCF7 cell lines, representing normal breast and breast cancer cells, respectively. The ultimate objective was to identify the most significant features that differentiate these cell lines. Materials and Methods: We utilized advanced imaging techniques such as holographic time-lapse microscopy, which provides real-time, three-dimensional imaging of cells, to conduct our comparative analysis. This allowed us to examine dynamic cellular morphology and behavior with exceptional sensitivity and resolution over time. The primary features assessed in our study included texture clustershade, area (μm2), eccentricity, irregularity, phaseshift sum, optical volume (μm3), shape convexity, and Hull convexity. Results: Our findings highlighted significant differences in the morphological features of MCF10A and MCF7 cells. MCF10A cells showed a higher texture clustershade value, suggesting less symmetry than MCF7 cells. On the other hand, MCF7 cells had smaller cellular area, higher eccentricity, lower irregularity, higher phase shift sum, higher optical volume, higher shape convexity, and higher hull convexity compared to MCF10A cells. These results suggest that MCF7 cells are smaller, more circular, less irregular, exhibit different light properties, and have a closer to perfect 3D shape relative to MCF10A cells. Conclusion: The identified morphological differences between MCF10A and MCF7 cells offer valuable insights into the characteristics distinguishing normal breast cells from breast cancer cells. These findings not only contribute to our understanding of the morphological variability in breast cancer but also underscore the potential utility of these differences in future cancer diagnostics and treatment strategies.
With its multitudinous complexity, cancer research constantly pushes the boundaries of our understanding of cellular and molecular processes that define the disease, fueling advancements in diagnosis, treatment, and overall disease management (1). A pivotal component of this research involves a detailed study of cancerous cells and their non-cancerous analogues, as differences between them may offer key insights into novel therapeutic interventions (2). Breast cancer, one of the most common cancers worldwide and a major cause of cancer-related mortality (3), is a heterogenous disease with diverse subtypes, each possessing unique biological characteristics. Thorough investigation of these cell lines’ morphological and behavioral traits can elucidate their proliferative and invasive potential (4).
The human breast cancer cell line, MCF7, and the non-tumorigenic human breast epithelial cell line, MCF10A, are widely used in such research endeavors. MCF7, due to its hormone sensitivity, is an important in vitro model for certain types of breast cancer (5), while MCF10A serves as a model for normal breast cells, offering a comparison point to cancerous cells (6). Notwithstanding the extensive research on these cell lines, the full extent of their morphological characteristics remains underexplored, leaving gaps in our comprehensive understanding. In this context, the application of advanced imaging techniques like holographic time-lapse microscopy is promising (7). This technique offers real-time, three-dimensional imaging of cells, enabling the study of dynamic cellular morphology and behavior over time with exceptional sensitivity and resolution.
Our study seeks to harness the power of holographic time-lapse microscopy for a meticulous comparison of MCF10A and MCF7 morphological features. The findings are expected to reveal critical differences between normal and cancerous breast cells, significantly enriching our understanding of breast cancer. An enhanced comprehension of these cellular differences could be instrumental in guiding future research and evolving treatment modalities for improved patient outcomes. By comprehensively elucidating the morphology of these cell lines, we could gain a deeper understanding of cellular processes, migration behaviors, and responses to therapeutics. The knowledge generated from this study could further our understanding of breast cancer cell dynamics, potentially leading to the development of innovative and targeted therapeutic strategies.
Materials and Methods
Cell culture. MCF10A cells were seeded in the 24-well plates 24 h before cells were visualized under a holomonitor. MCF10A cells were maintained in MEBM Basal Medium supplemented with BPE, hEGF, Insulin, hydrocortisone, GA-1000 (Lonza, Basel, Switzerland). MCF7 cells were maintained in DMEM medium (Gibco, New York City, NY, USA), 10% FBS (HyClone, Logan, UT, USA), 5% Penicillin/Streptomycin, Insulin (1 mg/ml, Gibco) at 37°C with 5% CO2.
Quantitative phase imaging. The digital holographic monitor investigations were carried out with a HoloMonitor M4, produced by Phase Holographic Imaging (PHI, Lund, Sweden), which was placed inside the mammalian cell culture incubator at 37°C with a 5% CO2 environment. Time-lapse phase imaging, alongside image processing, segmentation, and analysis were performed via the Hstudio software package provided by PHI. The proliferation of cells was evaluated by imaging every 15 min following 24 h after plating. In each well, the images were segmented, identified and cells were counted using the Cell Count module in HStudio software. For the tracking assays mentioned standard culture conditions were followed and cells were plated in 24-well plates (Sarstedt, Hildesheim, Germany). The motility of each cell was assessed at each time point by calculating the displacement of the object center between two consecutive images. Cells that were not tracked for the complete duration were omitted from the analysis. The displacement over time or total motility (the accumulation of all motilities over the imaging period) was used to calculate cell motility.
Box plots. We initiated our analysis by generating box plots for each morphological feature, separated by treatment group (MCF10A and MCF7). Box plots are a standardized way of displaying the distribution of data based on a five-number summary [“minimum”, first quartile (Q1), median, third quartile (Q3), and “maximum”]. It can tell us about our outliers and what their values are. It can also tell us if our data is symmetrical, how tightly it is grouped, and if and how it is skewed.
Violin plots. Following the box plots, we created violin plots for each morphological feature. Violin plots are similar to box plots, but they also show the probability density of the data at different values. These plots are decorated with a kernel density estimate to show the distribution shape of the data. The thick black bar in the center represents the interquartile range, the thin black line extended from it represents the 95% confidence intervals, and the white dot is the median. Violin plots are more informative than plain box plots or histograms because they provide a mirrored density plot.
Statistical analysis. The dataset was first loaded into a pandas DataFrame for ease of manipulation and analysis. The data consists of several morphological features of two cell lines, MCF10A and MCF7. We employed a feature selection technique based on mutual information to identify the features that most significantly differentiate between the two cell lines. This technique ranks the features based on the amount of information they provide about the cell line type. The top ten features were selected for further analysis. The top ten features were selected based on mutual information using a feature selection technique. Mutual information is a measure of the amount of information that knowing the value of one variable provides about another variable. In this case, we calculated the mutual information between each feature and the cell line type (MCF10A or MCF7). The mutual information measures the reduction in uncertainty about one variable given the knowledge of another. If the mutual information is high, then knowing the value of one variable significantly reduces the uncertainty about the other. In the context of this analysis, a high mutual information between a feature and the cell line type indicates that the feature provides a lot of information about the cell line type. Finally, the features were ranked by their mutual information with the cell line type, and the top ten features were selected for further analysis. This common approach in feature selection is used, as it allows us to focus on the most relevant features to the variable of interest.
To assess the statistical significance of the differences observed in these features between the two cell lines, we performed t-tests. The t-test is a statistical hypothesis test that determines whether there is a significant difference between the means of two groups. In this case, the two groups are the MCF10A and MCF7 cell lines, and the means are the mean values of the features for each cell line. The result of the t-test is a p-value, which is a measure of the probability that the observed difference occurred by chance. A p-value less than 0.05 is generally considered statistically significant.
In addition to the t-tests, we also generated boxplots, violin plots, and density plots for each of the top 10 features. These plots provide a visual representation of the distribution of each feature for the two cell lines. The boxplots show each feature’s median, quartiles, and potential outliers. The violin plots combine the boxplot with a kernel density estimation to provide a more detailed view of the distribution.
Finally, we calculated the mean values and effect sizes for each of the top 10 features. The mean values represent the average value of each feature for the two cell lines. The effect size (calculated here using Cohen’s d) is a measure of the magnitude of the difference between the two cell lines, in terms of standard deviation units. This provides a measure of the practical significance of the differences, complementing the p-values which provide a measure of statistical significance. The analysis was performed using Python, with the pandas library for data manipulation, the sklearn library for feature selection and t-tests, and the seaborn and matplotlib libraries for data visualization.
Results
In this study, we conducted a comprehensive analysis of morphological feature changes between MCF10A, a normal breast cell line, and MCF7, a breast cancer cell line (Figure 1). In our analysis of the breast cancer dataset, we aimed to identify the morphological features that most significantly differentiate between MCF10A and MCF7 cell lines. To achieve this, we employed a feature selection technique based on mutual information, followed by t-tests to assess the statistical significance of the differences observed. We also generated boxplots, violin plots, and density plots for each of the top 10 features to visually inspect the distributions of these features for the two cell lines.
Digital Holographic images of the (A). MCF10A and (B). MCF7 pseudo colored using HStudio.
Our analysis revealed the top features that provide the most information about the cell line type. In order of importance, these features are: texture cluster shade, area (μm2), eccentricity, irregularity, phaseshift sum, optical volume (μm3), shape convexity, and hull convexity.
We calculated the mean values of these features for the two cell lines and found that for Texture Clustershade, MCF10A to MCF7 was 4.5 vs. −2.9 (Figure 2). Texture cluster shade measures the image symmetry- the higher the cluster shade the less symmetrical the image is (8). Based on this, MCF10A cells seem less symmetrical then MCF7 cells.
Box plots and violin plots of the texture clustershade, area, eccentricity, irregularity, optical volume, and phaseshift sum. Images were taken with the PHI digital phase imaging microscope (PHI, Lund, Sweden). Images were processed and analyzed using Hstudio. A, B: Violin and box plots of texture clustershade; C, D: Violin and box plots of area; E, F: Violin and box plots of eccentricity; G, H: Violin and box plots of irregularity; I, J: Violin and box plots of Otical volume; K, L: phaseshift sum.
We also calculated the area and it was found that MCF10A and MCF7 had 552.9 vs. 478.9 μm2, respectively (Figure 2); for Eccentricity 0.66 vs. 0.58 (Figure 2); for Irregularity 0.37 vs. 0.29 (Figure 3); for phase shift sum 561 vs. 493 (Figure 2); for optical volume 2728 vs. 2397 μm3 (Figure 2); for shape convexity 0.88 vs. 0.91 (Figure 3); for hull convexity 0.93 vs. 0.94 (Table I). The mean ‘Texture clustershade’ for MCF10A cells was higher than that for MCF7 cells, indicating that MCF10A cells tend to have a higher texture clustershade. The ‘Texture clustershade’ feature emerged as the most informative, suggesting that variations in texture clustershade are strongly associated with the differences between MCF10A and MCF7 cell lines.
Box plots and violin plots of the shape and hull convexity. Images were taken with the PHI digital phase imaging microscope (PHI, Lund, Sweden). Images were processed and analyzed using Hstudio. A, B: Violin and box plots of shape convexity; C, D: Violin and box plots of Hull convexity.
Top 10 morphological features regarding the differences between MCF10A and MCF7 cells.
Similarly, area (μm2), which measures the size of the cell and eccentricity, which measures how much the cells deviates from a perfect circle (8), revealed that MCF7 cells have a smaller area and are more circular than MCF10A cells. This information can provide significant information about the cell line type indicating how area and eccentricity are associated with the cell line type. These findings provide valuable insights into the morphological characteristics that distinguish between MCF10A and MCF7 cell lines. Further studies are needed to validate these results and explore their potential implications for breast cancer diagnosis and treatment.
Another cellular feature that exhibited a change compared to healthy cells was the optical volume, where MCF7 cells exhibited a higher optical volume than MCF10A cells (Figure 2). Additionally, phaseshift sum, a measure of the light shift caused by the cells when the light passes through measured via digital holography, revealed that it was higher in MCF7 cells than MCF10A cells (8). Hull convexity, which measures the cells 3D shape was also observed to be significantly higher in MCF7 cells. The higher this number, the less dips in cell thickness and more perfect 3D cellular shape (8). So, MCF7 cells exhibit a higher hull convexity value, suggesting a cellular shape that is closer to a perfect 3D shape. Finally, irregularity is a measure of how much the circumference of the cell shape deviates from the perfect circle (8). Our data reveals that MCF7 cells decrease in irregularity compared to MCF10A cells. In conclusion, our analysis suggests significant morphological differences between the MCF10A healthy breast tissue cells and the MCF7 breast cancer cells. However, it is important to note that while our analysis provides valuable insights, further experimental studies would be needed to confirm these findings and fully understand the underlying biological mechanisms.
Discussion
The results of our study provide valuable insights into the morphological differences between the MCF10A normal breast cell line and the MCF7 breast cancer cell line. Through a comprehensive analysis of morphological features, we identified several key characteristics that differentiate the two cell lines.
One of the most informative features was the texture clustershade, which measures the image symmetry. We found that MCF10A cells exhibited a higher texture clustershade value than MCF7 cells, indicating that MCF10A cells tend to be less symmetrical. This finding suggests that symmetry variations play a role in distinguishing between normal and cancerous breast cells.
Additionally, features such as area, eccentricity, irregularity, phase shift sum, optical volume, shape convexity, and hull convexity were significantly different between the two cell lines. MCF7 cells showed smaller cellular area, higher eccentricity, lower irregularity, higher phase shift sum, higher optical volume, higher shape convexity, and higher hull convexity compared to MCF10A cells. These findings suggest that MCF7 cells are smaller, more circular, less irregular, exhibit different light properties, and have a closer to perfect 3D shape compared to MCF10A cells.
The observed differences in these morphological features provide important insights into the characteristics that distinguish normal breast cells from breast cancer cells. Further research and validation studies are warranted to confirm these findings and explore their potential implications for breast cancer diagnosis and treatment. It is important to acknowledge the limitations of our study. The analysis was based on the specific MCF10A and MCF7 cell lines, and the findings may not be generalizable to other breast cell lines or cancer subtypes. Additionally, morphological features alone may not be sufficient for accurate cancer diagnosis, and additional molecular or genetic analysis is usually required.
In conclusion, our study highlights the significance of morphological features in differentiating between normal and cancerous breast cells. The identified differences in texture cluster shade, area, eccentricity, irregularity, phase shift sum, optical volume, shape convexity, and hull convexity provide valuable insights into the distinctive characteristics of MCF10A and MCF7 cell lines. These findings contribute to our understanding of breast cancer morphology and may have implications for future research in cancer diagnostics and treatment strategies.
Footnotes
Authors’ Contributions
Study conception and design: B.X. Acquisition of data: B.X., A.J.M., A.J.K., D.I.M., Analysis and interpretation of data: B.X., A.J.M., A.J.K., D.I.M. Drafting of the manuscript: B.X. The Authors read and approved the final manuscript.
Conflicts of Interest
The Authors declare no competing interests.
Funding
This work was supported by the University of Michigan – Dearborn, the University of Michigan – Dearborn B. X. Start-Up funds, and the University of Michigan – Dearborn Research Initiation Development Funds.
- Received June 15, 2023.
- Revision received July 17, 2023.
- Accepted July 21, 2023.
- Copyright © 2023 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.