A quantitative proteomics analysis of MCF7 breast cancer stem and progenitor cell populations

Abstract

Accumulating evidence has demonstrated that breast cancers are initiated and develop from a small population of stem-like cells termed cancer stem cells (CSCs). These cells are hypothesized to mediate tumor metastasis and contribute to therapeutic resistance. However, the molecular regulatory networks responsible for maintaining CSCs in an undifferentiated state have yet to be elucidated. In this study, we used CSC markers to isolate pure breast CSCs fractions (ALDH+ and CD44+CD24- cell populations) and the mature luminal cells (CD49f-EpCAM+) from the MCF7 cell line. Proteomic analysis was performed on these samples and a total of 3304 proteins were identified. A label-free quantitative method was applied to analyze differentially expressed proteins. Using the criteria of greater than twofold changes and p value <0.05, 305, 322 and 98 proteins were identified as significantly different in three pairwise comparisons of ALDH+ versus CD44+CD24-, ALDH+ versus CD49f-EpCAM+ and CD44+CD24- versus CD49f-EpCAM+, respectively. Pathway analysis of differentially expressed proteins by Ingenuity Pathway Analysis (IPA) revealed potential molecular regulatory networks that may regulate CSCs. Selected differential proteins were validated by Western blot assay and immunohistochemical staining. The use of proteomics analysis may increase our understanding of the underlying molecular mechanisms of breast CSCs. This may be of importance in the future development of anti-CSC therapeutics.

Keywords: Breast cancer; Cancer stem cell; LC-MS/MS; Pathway analysis; Quantitative proteomics; Systems biology.

Figure 1

Schematic of experimental workflow. ALDH+, CD44+CD24− and CD49f-EpCAM+ cell populations were sorted from MCF7 cells. Then the extracted proteins from about 50 000 cells were digested by the FASP method. LC-MS/MS based on Orbitrap Elite MS and spectral counting methods were applied to analyze and quantify differentially expressed proteins among the three different fractions.

Figure 2

Flow cytometry sorting process to isolate MCF7 CSCs and mature luminal cells. Left) CD44+CD24− cells were gated (P1) and sequentially gated for ALDH− (P2) and not CD49f-EpCAM+ (P3) cells. Middle) ALDH+ cells were gated (P4) and sequentially gated for not-CD44+CD24− (P5) and not CD49f-EpCAM+ (P6) cells. Right) CD49f-EpCAM+ cells were gated (P7) and sequentially gated for not-CD44+CD24− (P8) and ALDH− (P9) cells. The frequencies represent values based on each parental gate.

Figure 3

Proteome analysis of identified results. (A) The overlap of proteins identified from ALDH+, CD44+CD24− stem cell populations and CD49f− EpCAM+ non-stem cell population. Comparison of Gene Ontology annotations for cellular component (B) and biological process (C) of proteins identified among the three cell populations using DAVID bioinformatics database.

Figure 4

(A) Histograms of log₂-transformed protein ratios for pairwise comparisons. The ratios for ALDH+/CD44+CD24− and ALDH+/CD49f-EpCAM+ show larger deviation than those for CD44+CD24/CD49f-EpCAM+ comparison. (B) Volcano plot of ratios and p-value of the three pairwise comparisons. Proteins with p< 0.05 and above/below twofold changes are identified as proteins with significant changes. (C) Hierarchical cluster analysis of differentially expressed proteins among the three samples. The rows represent each protein and the column means different runs. Three replicates were clustered together. The clustering method for x and y axis used here is K-means clustering. Euclidean method was used to calculate distance.

Figure 5

Canonical signaling pathways enriched in differentially expressed proteins among ALDH+, CD44+CD24− stem cell populations and CD49f-EpCAM+ non-stem cell populations. A p-value threshold of 0.05 is applied (dotted line).

Figure 6

Top biological network generated by Ingenuity Pathway Analysis for differentially expressed proteins of (A) ALDH+/CD44+CD24−, (B) ALDH+/CD49f-EpCAM+ and (C) CD44+CD24−/CD49f-EpCAM+. Red and green represent over- and underexpressed proteins identified in this work; white indicates proteins that were not in the differentially expressed protein lists but related to the network. Solid arrows represent known direct interactions, and dotted arrows mean indirect interactions.

Figure 7

(A) Western blotting analysis of HSPB1 and ANXA3 in the three cell samples. Beta-actin served as control. (B) Immunohistochemical staining analysis of ANXA3 in normal and breast cancer tissue microarrays. A representative figure is presented (Magnification 20×). In normal breast tissue, ANXA3 had weak, medium or strong expression level in 73%, 21% and 6% of samples, respectively. In breast cancer samples, ANXA3 had weak, medium or strong expression level in 20%, 32% and 48% of samples, respectively.