Dissecting Morphological and Functional Dynamics of Non-Tumorigenic and Triple-Negative Breast Cancer Cell Lines Using PCA and t-SNE Analysis

Abstract

Background: Triple-negative breast cancer (TNBC) poses significant challenges due to its aggressive nature and lack of targeted therapies. Understanding the cellular behaviors of TNBC is crucial for developing effective treatments.

Aims: This study aims to compare the morphological characteristics of non-tumorigenic MCF10A and aggressive MDA-MB-231 TNBC cell lines using advanced analytical techniques.

Methods and results: Advanced techniques such as Principal Component Analysis (PCA), t-Distributed Stochastic Neighbor Embedding (t-SNE), and digital holographic microscopy were utilized. Cellular features such as area, migration, motility, irregularity, and optical thickness were thoroughly analyzed over time. Our results revealed significant morphological differences between the MCF10A and MDA-MB-231 cell lines. Specifically, MDA-MB-231 cells displayed enhanced motility and a smaller, more variable size, attributes that may facilitate their invasive potential. In contrast, MCF10A cells exhibited larger sizes and more regular migration patterns, suggesting stability in structured tissue environments. Additionally, temporal analysis highlighted consistent phenotypic behaviors over time, with MDA-MB-231 cells demonstrating higher optical thickness and irregularity, indicating potential structural complexities associated with malignant transformation. Correlative analysis further confirmed these results by revealing connections between cell size, motility, and optical properties crucial for understanding cell behavior within their microenvironment.

Conclusion: The profound differences in cellular dynamics between MCF10A and MDA-MB-231 cell lines underscore the unique adaptive mechanisms of TNBC cells. Our study provides valuable insights into the cellular foundations of TNBC aggressiveness, offering a foundation for future research aimed at understanding the mechanistic underpinnings of TNBC progression and therapeutic targeting.

Keywords: MCF10A; MDA‐MB‐231; PCA; breast cancer; cell motility; digital holographic microscopy; temporal dynamics; triple‐negative breast cancer; t‐SNE.

FIGURE 1

Digital Holographic Microscopy (DHM) images of MCF10A and MDA‐MB‐231 cell lines. The 3D topographical representation of cellular morphology is shown, with color‐coded height maps indicating optical thickness (in micrometers). Warmer colors (red and yellow) correspond to areas of higher optical thickness, while cooler colors (blue) indicate lower optical thickness. The left panel represents the non‐tumorigenic MCF10A cells, while the right panel shows the triple‐negative breast cancer MDA‐MB‐231 cells. The scale bars represent 100 μm. These images highlight the morphological and structural differences between the cell lines. These are raw phase images acquired via label‐free digital holographic microscopy (DHM); cell boundaries appear subtle due to the nature of this imaging modality.

FIGURE 2

PCA and t‐SNE plots of MCF10A and MDA‐MB‐231 cells. Each point represents a cell, with black points representing MCF10A cells and red points representing MDA‐MB‐231 cells. (A) In the PCA plot the x and y axes represent the first two principal components. (B) The t‐SNE plot of MCF10A and MDA‐MB‐231 cells. Each point represents a cell, with black points representing MCF10A cells and red points representing MDA‐MB‐231 cells. The x and y axes represent the two t‐SNE dimensions.

FIGURE 3

Box plots for six selected features (Area, Migration, Motility, Irregularity, Migration directness, and Optical thickness avg). Each box plot presents the distribution of a feature for the two cell lines, with black representing MCF10A and red representing MDA‐MB‐231. Statistical significance is indicated by p‐values, and effect sizes are shown as Cohen's d values.

FIGURE 4

Line plots over time for six selected features. Each plot presents the temporal dynamics of a feature for the two cell lines, with black representing MCF10A and red representing MDA‐MB‐231. The x‐axis represents time in minutes, and the y‐axis represents normalized feature values. (A) Avg. Area (μm²), (B) Avg. Migration (μm), (C) Avg. Motility (μm), (D) Avg. Irregularity, (E) Avg. Migration directness, and (F) Avg. Optical thickness avg. (μm) over time.

FIGURE 5

Analysis of morphological feature relationships. (A) Heatmap of correlation matrix. The heatmap presents the correlations between different features, with color intensity and hue representing the strength and direction of the correlation. Positive correlations are represented in red, and negative correlations are represented in blue. (B) Hierarchical clustering dendrogram of morphological features. The dendrogram illustrates the relationships between features based on similarity, with features that are more closely related positioned nearer to each other on the branching structure.

FIGURE 6

Scatter plot matrix illustrating pairwise relationships among cellular features for MCF10A (black) and MDA‐MB‐231 (red) cells: Average area (μm²), average irregularity, average migration (μm), average migration directness, average motility (μm), average motility speed (μm/h), average optical thickness (μm), average optical volume (μm³), and centroid X and Y positions. The diagonal panels show the distribution of each feature for both cell lines. Off‐diagonal panels display scatter plots for every possible feature pair, overlaid with a smoothed trend line. In the upper right corner of each scatter panel, three values are provided: The overall Pearson correlation coefficient (r) for all cells, the cell‐line–specific r values for MCF10A and MDA‐MB‐231, and the statistical significance of the overall correlation (*p < 0.05; **p < 0.01; ***p < 0.001).

FIGURE 7

Forest plot of Cohen's d effect sizes for six key features. Each point represents the effect size (Cohen's d) comparing MCF10A and MDA‐MB‐231 cells; error bars denote ±1 SE, and asterisks indicate statistical significance (*p < 0.1;***p < 0.001). Black squares (■) indicate features higher in MCF10A; red circles (●) indicate features higher in MDA‐MB‐231.