Introduction
Epithelial to mesenchymal transition has been shown to be a necessary step in the process of forming metastasis [
1,
2]. In order to characterize these cells in the background of a single genome, an epithelial clone of PC3 cells, PC3-Epi, was isolated along with a mesenchymal derivative of PC3 known as PC3-EMT. Although PC3-Epi and PC3-EMT have similar growth rates when grown in subcutaneous mouse xenografts, PC3-EMT had a fourfold higher rate in their ability to form multiple metastatic lesions after mouse intracardial injection. Additionally, both cell lines undergo EMT and the reverse process of MET within these sites of metastasis as evidenced by ZEB1 (mesenchymal) and E-cadherin (epithelial) immunohistochemical staining [
3]. In order to isolate epithelial and mesenchymal prostate cells within the patient bone marrow without fixing them, flow cytometric methods need to be developed. As stated, ZEB1 and E-cadherin are excellent markers; however, ZEB1 requires fixation of cells and E-cadherin is not specific enough to use as a stand-alone marker for metastatic prostate cancer. Here, we will demonstrate that RNA-seq of the PC3-Epi and PC3-EMT cell lines suggests that CD44 would be a good candidate.
The hyaluronic acid receptor CD44 and its many isoforms are associated with a wide variety of cell types including epithelial, mesenchymal and CSCs (cancer stem cells). Previous studies have also shown that these isoforms are related to various survival functions and tumorigenesis [
4]. Therefore, in order to differentiate between cell types, it is critical to understand and distinguish among the numerous isoforms of CD44. The CD44 gene consists of 20 exons and is subject to significant alternative splicing [
5]. Within the extracellular domain of CD44, there exists a highly variable 10 exon region and it is the splicing pattern of these 10 exons that defines a given isoform. Typically, CD44 is categorized into one of two groups: those expressing different combinations of these 10 variant exons (CD44v) and those that do not (CD44s). Each group has been demonstrated to be linked to different functions, pathways and cell types. CD44s is typically associated with mesenchymal stem cells, while the various CD44v forms are related to hematopoietic stem cells and increased cell adhesion. However, both CD44s and CD44v have been linked to the CSC niche and cancer progression [
6,
7]. The epithelial to mesenchymal transition (EMT) process whereby cancer cells lose polarity, cell–cell adhesion and gain an invasive capacity is essential for metastasis in a wide range of cancers [
8]. CD44 splicing from variant (CD44v) to the standard (CD44s) form has been shown to be essential for an EMT in breast cancer [
9,
10]. This up-regulation in the CD44s isoform is caused by decreased expression of epithelial splicing regulatory protein 1 (ESRP1). ESRP1 is the splicing factor for variant forms of CD44; it splices CD44 by binding to intronic regions. By itself, expression of ESRP1 prevents EMT [
11]. Despite the knowledge of the various isoforms and their unique functions, total CD44 expression is most often used as a surface marker for cell stemness rather than assaying for any specific variant isoform [
12].
Previously, our laboratory demonstrated that epithelial cancer cells undergo an EMT upon exposure to M2 macrophages [
3,
13]. In order to obtain a purely epithelial PCa population, our laboratory isolated a single cell clone of PC3 that had high E-cadherin and low vimentin expression, denoted PC3-Epi. Alternatively, PC3 cells were incubated with M2 macrophages, which caused a stable EMT to occur after only a few days in culture and were denoted PC3-EMT. The EMT process was found to necessitate the down-regulation of the OVOL1 and OVOL2 transcription factors and up-regulation of ZEB1. Additionally, ESRP1 was most highly positively correlated with the expression of OVOL1/2 (
r = 0.76, 0.84 respectively) in a series of 917 cancer cell lines. All this indicates that CD44 splicing maybe a tractable cell surface marker for differentiating epithelial from mesenchymal cells.
Discussion
In this work, we discovered that the expression of CD44 variant exon 6 (CD44v6) is greater in PC3-Epi than PC3-EMT. It was also very recently shown that CD44v6 expression is necessary for anchorage independent growth and that high expression is associated with a poor outcome in prostate cancer patients [
21]. Additionally, other studies in lung, colon and breast indicate that stronger CD44v6 expression is a negative prognostic indicator [
22‐
24]. In order to determine the relationship between EMT, CD44 isoforms and ALDH, we characterized PC3-Epi and PC3-EMT via qPCR, FACs and western blotting.
Using FACs, the CD44v6 variant was more highly enriched at the protein level in PC3-Epi than in PC3-EMT, but expression of CD44v4/7 was much less dramatic. This is somewhat surprising since the isoforms present in PC3-Epi/EMT (Fig.
3a) always have v4/v6/v7 coexpressed in the same isoforms (Fig.
3a). It is possible that the antibodies have a difficult time recognizing CD44v4 and CD44v7 as an epitope, possibly due to glycosylation or some other post-translational modifications that mask antibody binding sites [
25,
26]. It is surprising that PC3-EMT elicits any CD44v6 expression since it is undetectable at the western and qPCR levels. In all likelihood, PC3-EMT will be shown to be null by flow cytometry when more specific antibodies are available.
Immunohistochemistry from various groups is inconclusive about the value of CD44v6 as a predictor for prostate cancer survival [
27‐
29]. Given our FACs results, this is not surprising that immunohistochemistry only relies on immunoaffinity [
30]. FACs lacks the second dimension that a gel sizing can add which reduces the specificity of flow cytometry. Our results demonstrate that many PC3-EMT have CD44v6; however, they clearly do not according to qPCR. Additionally, many PC3-Epi cells are negative for CD44v7 which is very unlikely given the sequence of the isoforms shown in the RNAseq data all have CD44v7 coexpressed with CD44v6.
Interestingly, it has been suggested that CD44v6 is important in drug resistance since knock down of CD44v6 prevented colony formation after exposure to various chemotherapeutics. Somewhat unexpectedly, however, knock down of CD44v6 also diminished the expression of mesenchymal markers, including vimentin, Snai1/2 and Twist. The authors suggest that this is due to decreased AKT activity with CD44v6 being a constituent of the WNT signaling pathway, a known inducer of EMT. This seems to argue with our data since all PC3-Epi expressed CD44v6, while only 50 % of PC3-EMT was CD44v6 positive by flow cytometry. However, it may be that CD44v6 is needed at low levels to stimulate WNT but at higher levels or when coexpressed with other variants it functions in alternative pathways, which do not induce EMT [
21,
31].
In conclusion, once specific antibodies are made to either CD44v6 or one of the other variant exons, it is likely that it will be an excellent way to differentiate cells by flow cytometry. In the future, the establishment of antibodies needs to better take into account post-translational modifications, such as N/O-linked glycosylation, which are differentially utilized in mesenchymal vs. epithelial cells [
32]. There has been much work that has shown that CD44v expression alone prevents EMT [
9,
33], which means that CD44 splicing alone maybe just as good, if not better than cell adhesion proteins such as E-cadherin in predicting the differentiation status of cells. Unlike E-cadherin, which was only 40 % positive in a pure population, CD44v6 was 100 % positive in the PC3-Epi line, indicating that it may be a superior method to distinguish epithelial cells.
Acknowledgments
This work was supported by NCI Grant Nos. U54CA143803, CA163124, CA093900 and CA143055.to K.J.P.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.