CD44+/CD24-phenotype of breast cancer cells is associated with invasive properties
We investigated the importance of the stem/progenitor phenotype defined by CD44 positivity and CD24 negativity for breast cancer cells to invade and metastasize. Metastasis is a complex process that involves integrated activity of genes, which function in discrete steps that include the following: angiogenesis, invasion, intravasation, survival in circulation, extravasation, and homing and proliferation at sites of metastasis [
19,
34]. These genes include UPA/UPA receptor, MMPs, cytokines such as IL-1, IL-6, IL-8 and IL-11, parathyroid hormone-related peptide and the chemokine receptor CXCR4 [
12,
14,
21,
31]. Here we show that several of these genes are expressed in cell lines that contain significant numbers of CD44
+/CD24
- cells and that the expression pattern of these genes and the CD44
+/CD24
- phenotype correlates with invasive behavior of cell lines. However, the CD44
+/CD24
- phenotype is not sufficient for homing and growth at sites of metastasis. Thus, steps in the cascade of events required for the spread of cancer are dependent on distinct groups of genes, and the CD44
+/CD24
- phenotype may define the expression of the group of genes involved in invasion.
CD44 and CD24 have been shown to regulate invasion and metastasis of breast cancer cells either positively or negatively. Although most studies have shown CD44-mediated invasion of breast cancer cells [
35,
36], Lopez and coworkers [
37] showed inhibition of breast cancer metastasis by this molecule. Similarly, CD24 has been shown to promote [
38] or inhibit [
39] invasion and metastasis of breast cancer cells. However, association of CD44
+/CD24
- phenotype with invasion observed in this study is not linked to the function of these genes because MDA-MB-468 cells expressing both CD44 and CD24 failed to invade. Also, the CD44
+/CD24
- subpopulation of TMD-436 was more invasive than the CD44
+/CD24
+ subpopulation of the same cell line (Figure
3b). Thus, the invasive property is intrinsic to cells of CD44
+/CD24
- phenotype. Which among the genes expressed in CD44
+/CD24
- cells confers the invasive phenotype is yet to be determined because CD44
+/CD24
- and CD44
+/CD24
+ subpopulations of TMD-436 cells exhibited modest differences in expression levels of the proinvasive genes tested but exhibited differences in invasion. In the original study on tumorigenic breast cancer progenitor cells [
9] CD10 and CD140b were used as lineage markers, and so cancer cells expressing CD10 or CD140b were excluded from progenitor cells. However, several breast cancer cell lines that we examined express CD10 and it is considered a basal cell marker [
25]. Studies are underway to determine whether the CD44
+/CD24
- subpopulation can be further subdivided based on the expression of markers such as CD10 and whether such subsets have unique invasion/metastasis properties.
The invasive metastasis properties of several of the cell lines that we used in this study were examined by others before the identification of CD44
+/CD24
- cells, and most of these studies correlated invasive properties with ER-α status and/or expression status of mesenchymal markers such as vimentin or MMPs [
24,
40,
41]. These studies established a general trend toward increased invasiveness of ER-α-negative breast cancer cells. However, not all ER-α-negative cells were invasive (MDA-MB-468 and SK-BR-3 cells, for example) [
24,
42]. Our study clearly shows a direct association between the CD44
+/CD24
- phenotype and invasion. However, neither our study nor previous studies revealed an association between homing and proliferation at sites of metastasis and ER-α status, mesenchymal marker expression, or CD44
+/CD24
- phenotype. For example, the ER-α-positive and vimentin-negative cell line MCF-7 lacking CD44
+/CD24
- subpopulation forms osteosclerotic bone lesions on intracardiac injection in nude mice [
32]. Similarly, we observed lung metastasis of MDA-MB-468 cells, which are ER-α negative and vimentin negative, and lack a CD44
+/CD24
- subpopulation. In contrast, the vimentin-positive and ER-α-negative cell line Hs578T, which has an 86% CD44
+/CD24
- subpopulation, failed to form lung metastasis. The minor difference in our results and previously published data with respect to hematogenous metastasis of Hs578T on mammary fat pad injection [
24] is probably due to low frequency of metastasis (10%). Also, we did not observe metastasis of SUM1315 cells, with a 97% CD44
+/CD24
- subpopulation, in nude mice, although previous studies have shown bone metastasis of these cells in nonobese diabetic/severe combined immunodeficiency mice with humanized but not mouse bone [
43].
In a complementary study, Abraham and coworkers [
44] reported that the prevalence of CD44
+/CD24
- cells (tumors with >10% of CD44
+/CD24
- cancer cells) in 22% of tumor samples. The prevalence of these cells correlated with distant metastasis but no other clinical parameters. Our data suggest that the CD44
+/CD24
- population plays a critical role in the invasive step of metastasis. Thus, distant metastasis in patients with elevated levels of CD44
+/CD24
- cells may be related to enhanced invasiveness of cancer cells. It is possible that the establishment of growth at sites of metastasis is controlled by a distinct set of genes whose expression is unrelated to the CD44
+/CD24
- phenotype. An emerging opinion is that reduced expression of genes involved in cell-cell communication initiates invasion, whereas re-expression of genes involved in cell-cell communication is essential for survival and reattachment of metastases [
45]. Therefore, metastatic growth may be primarily determined by signaling pathways that control the re-expression of cell-cell communication genes, which may be further influenced by the organ-specific microenvironment. In this regard, the transforming growth factor-β-activated signaling pathway is suggested to play a significant role in growth of cancer cells at sites of metastasis [
13,
14]. Although previous studies have identified lung metastasis signature genes using MDA-MB-231 cells as a model system [
30], our studies reveal that the same set of lung metastasis signature genes is not involved in metastasis of MDA-MB-468 cells. Thus, additional studies are required to elucidate the mechanisms of tumor cell growth at sites of metastasis.
CD44+/CD24-phenotype may define breast cancers of basal/myoepithelial origin
Molecular profiling studies have classified breast cancers to five types with distinct prognostic significance: luminal type A, luminal type B, ErbB2-positive, normal-like, and basal type [
28,
29]. Patients with luminal type A tumors have the most favorable prognosis, whereas patients with basal-type tumors have worst prognosis. Breast cancer cell lines have also been classified into five groups – luminal, basal, mesenchymal, ErbB2-positive, and myoepithelial – based on gene expression profiling [
25,
27]. As per gene expression profiling, basal and mesenchymal cells are similar except for differential expression of 227 genes. Interestingly, all cell lines that contained CD44
+/CD24
- population are in the basal/mesenchymal or the myoepithelial group (Table
1). Thus, stem/progenitor cells for luminal and ErbB2-positive breast cancers, which represent the majority of breast cancers, remain to be identified. In this regard, a 'side population' of cells with high drug efflux capacity has been described as cancer stem cells for breast cancer, lung cancer, and glioblastoma [
46,
47]. These side population cells have been identified in breast cancer cell lines of luminal type, and these cells overexpress transporter genes ABCG2 (ATP-binding cassette, subfamily G, member 2) and ABCA3 (ATP-binding cassette, subfamily A, member 3) [
46]. Cells that express higher levels of CD24 in mouse are defined as luminal epithelial cells, and Lin
-CD29
hiCD24
+ cells have been defined as mammary stem cells in mouse [
48,
49]. Note that all luminal cell types contain disproportionately higher levels of CD44
-/CD24
+ cells (Table
1), and this population may contain cancer progenitor cells corresponding to luminal type of tumors. Identification of cancer stem cells specific for luminal cells, which represent about 70% of breast cancers, may allow improved understanding of signaling events that are involved in discrete steps of breast cancer progression, including metastasis.