CD44 staining of cancer stem-like cells is influenced by down-regulation of CD44 variant isoforms and up-regulation of the standard CD44 isoform in the population of cells that have undergone epithelial-to-mesenchymal transition

Abstract

CD44 is commonly used as a cell surface marker of cancer stem-like cells in epithelial tumours, and we have previously demonstrated the existence of two different CD44(high) cancer stem-like cell populations in squamous cell carcinoma, one having undergone epithelial-to-mesenchymal transition and the other maintaining an epithelial phenotype. Alternative splicing of CD44 variant exons generates a great many isoforms, and it is not known which isoforms are expressed on the surface of the two different cancer stem-like cell phenotypes. Here, we demonstrate that cancer stem-like cells with an epithelial phenotype predominantly express isoforms containing the variant exons, whereas the cancer stem-like cells that have undergone an epithelial-to-mesenchymal transition down-regulate these variant isoforms and up-regulate expression of the standard CD44 isoform that contains no variant exons. In addition, we find that enzymatic treatments used to dissociate cells from tissue culture or fresh tumour specimens cause destruction of variant CD44 isoforms at the cell surface whereas expression of the standard CD44 isoform is preserved. This results in enrichment within the CD44(high) population of cancer stem-like cells that have undergone an epithelial-to-mesenchymal transition and depletion from the CD44(high) population of cancer stem-like cells that maintain an epithelial phenotype, and therefore greatly effects the characteristics of any cancer stem-like cell population isolated based on expression of CD44. As well as effecting the CD44(high) population, enzymatic treatment also reduces the percentage of the total epithelial cancer cell population staining CD44-positive, with potential implications for studies that aim to use CD44-positive staining as a prognostic indicator. Analyses of the properties of cancer stem-like cells are largely dependent on the ability to accurately identify and assay these populations. It is therefore critical that consideration be given to use of multiple cancer stem-like cell markers and suitable procedures for cell isolation in order that the correct populations are assayed.

Figure 1. CD44 standard and variant exons.

A representation of the 9 standard and 9 variant exons of CD44, showing the names of each variant exon and where they fall in the extracellular region of the protein product. The position of the “Epitope 1” region recognised by the CD44 antibody used in these studies is indicated.

Figure 2. CD44 expression in EMT and non-EMT CSCs.

(A) FACS sorting of CA1, Met1 and Met2 cells by expression of CD44 (using the “epitope 1” antibody that detects all forms of CD44) and ESA. The CD44^highESA^low and CD44^highESA^high sub-populations are gated and show higher levels of staining of CD44^highESA^low cells for CD44. (B) QPCR analysis of CD44 variant gene expression in CD44^highESA^low cells relative to the CD44^highESA^high cells for the CA1, Met1 and Met2 lines shows marked differences in the isoform expression patterns.

Figure 3. Trypsin treatment prevents cell surface detection of CD44 variant isoforms.

Histograms showing expression, assessed by FACS analyses of CA1 cells, of CD44 (“epitope 1” antibody, top left panel) and of CD44 variants after treatment with trypsin (green) or enzyme-free buffer (blue). The isotype control is in red.

Figure 4. Detection of CD44 variant isoform expression on the CD44^highESA^high cell sub-population.

FACS analysis of CD44 variant isoform expression on cells of the CA1 (A), Met1 (B) and Met2 (C) cell lines isolated using enzyme-free buffer showing the CD44 variants on the y-axis and ESA expression on the x-axis. For all cell lines, similar total levels of CD44 are detected on the CD44^highESA^low and CD44^highESA^high cells but lower levels of the v3, v5, v6 and v9 isoforms are detected on the CD44^highESA^low cells. (D) FACS analysis of CD44 (y-axis) and ESA (x-axis) expression on CA1 (top), Met1 (middle) and Met2 (bottom) cells treated with either trypsin (left) or enzyme-free buffer (right). Loss of detection of CD44 variant isoforms after trypsin treatment results in a large increase in the fraction of motile CSCs within the 5% of cells showing the highest CD44 expression. Representative plots. (E) Quantification of the experiments depicted in figure 4D, showing the percentage of EMT CSCs within the 5% CD44^high population after treatment with trypsin or enzyme-free buffer.

Figure 5. Enzyme treatment decreases the size of the CD44-positive population.

Following isolation with trypsin, Accutase or enzyme-free (EF) buffer, CA1 cells were subject to FACS analysis either unstained, stained with an isotype control, or stained for CD44. Unstained and isotype-stained cells were unaffected by method of isolation but marked differences in the proportion of cells classified as staining positive for CD44 are seen.

Figure 6. Differential staining for CD44 on cells isolated from fresh tumour samples using different methods.

Panels A–D illustrate the procedure used to identify staining patterns. Cell smears were prepared and co-stained with DAPI (A) with a pan-keratin antibody (B) and with the CD44 “epitope 1” antibody (C). These panels show images manipulated by adjusting capture thresholds in PhotoShop to identify cells with set levels of staining above a background level. For each field, this level was held constant to generate the three images, one for each fluorochrome, containing only the stained cells. These images were then combined, slightly out of lateral register, to allow identification of cells with absent or low levels staining for CD44 or keratin (blue nuclei only), staining for CD44 only (CD44+), keratin only (K+), or both CD44 and keratin (CD44+K+). Treatment with trypsin+collagenase or collagenase alone consistently reduced the fraction of cells that was positive for CD44 (E) and for both CD44 and keratin (F). Displaying the 5 assayed tumours individually (G) demonstrates the striking variation in CD44 staining between tumours, and that this variation is reduced following enzymatic treatment. Statistical analysis was conducted using a Mann-Whitney test.