The pentaspan transmembrane glycoprotein CD133, also known as Prominin-1, was originally described as a hematopoietic stem cell marker [24
] and subsequently shown to be expressed by a number of progenitor cells, including those of the epithelium where it is expressed on the apical surface [25
]. Since the discovery of CD133+ brain tumor stem cells [5
], CD133 has been used as a marker for purifying cancer stem cells in other solid tumors, including liver [26
], pancreas [27
], prostate [8
] and colon [29
]. Significantly, a recent study of 8 primary ESFT reported the frequency of CD133+ tumor cells to be 4-8% and functional studies further implicated these cells as putative tumor-initiating cancer stem cells [16
We evaluated expression of the CD133-encoding gene PROM1 in a large cohort of primary ESFT. Consistent with an overall low frequency of CD133+ cells, we found PROM1 expression to be extremely low in most cases of ESFT. The absence of detection of PROM1 transcript in 5 tumors may be indicative of true negative status or, more likely, a reflection of the RNA having been isolated from very small, closed needle biopsy specimens. Despite the generally low levels of PROM1 detection in primary tumors, however, in a significant minority of cases (11 of 48) the transcript was readily detected. Intriguingly, EWS-ERG fusion positive cases were significantly over-represented among PROM1 expressing tumors. This was corroborated by in vitro studies, which also revealed levels of PROM1/CD133 to be, in general, higher in EWS-ERG than EWS-FLI1 cell lines. Whether this difference between tumors with different fusion types is a function of differential effects of the ETS fusion partner or a reflection of different cellular origins of EWS-FLI1 and EWS-ERG tumors is an intriguing question that will require further study.
Significantly, in two cases (one EWS-FLI1 and one EWS-ERG), high levels of PROM1 expression were associated with primary drug-resistant disease. Moreover, in one of these cases the frequency of CD133+ cells increased post-treatment suggesting that the CD133+ fraction contributed to treatment failure. In contrast, however, two other tumors with high levels of PROM1 responded well to standard therapy and both patients are long-term, event-free survivors. Unfortunately, frozen tissue was only available for one of the four PROM1-high cases so it is not known if high PROM1 levels in the other cases were also associated with high levels of glycosylated CD133. It may be that high levels of PROM1 are predictive of chemoresistant ESFT but only when accompanied by high expression of the glycosylated CD133 protein. Studies designed to simultaneously evaluate transcript expression levels as well as glycosylated and non-glycosylated CD133 protein expression in fresh frozen ESFT specimens are necessary to address this issue. Thus, expression of PROM1 in ESFT is highly variable and the potential clinical significance of high level PROM1 transcript expression requires further evaluation in large, prospective studies.
The existence of discrete populations of tumor-initiating cells within established cultures indicates that even cell lines that have been maintained for many years in culture can retain cellular hierarchies in which tumorigenic stem cells give rise to less tumorigenic progeny [15
]. Functional studies of CD133+ and CD133- fractions derived from ESFT cell lines demonstrated significant heterogeneity in their biologic properties. Specifically, although CD133+ cells could be isolated from all ESFT cell lines, only CD133+ cells isolated from the STA-ET-8.2 cell line displayed evidence of stem cell characteristics and chemo-resistance. In contrast, we could discern no phenotypic or functional differences between CD133+ and CD133- cells derived from other ESFT cell lines. Whether this inconsistency is a result of genetic evolution in vitro
or a true reflection of variable significance of CD133 expression in the original tumors remains to be determined.
Importantly, recent studies have challenged the utility of CD133 as a single marker of tumor-initiating cell populations. CD133- tumor cells derived from some primary tumors and cell lines possess self-renewal and tumor-initiating potential even when injected at very low numbers [22
]. These conflicting data, when combined with the uncertain biological role of CD133, highlight the need for additional distinguishing markers that are directly involved in maintaining the functional properties of the putative cancer stem cell population [35
]. Indeed, the variability in PROM1
expression in primary ESFT combined with the inconsistent biologic properties of CD133+ ESFT cells in culture suggest that CD133 expression alone will be insufficient to isolate drug-resistant cancer stem cells in ESFT. Nevertheless, our data indicate that PROM1/
CD133 expression may be a useful marker of increased chemoresistance in at least some cases of primary ESFT and that the STA-ET-8.2 cell line will be a useful tool to study the biology of these cells in the laboratory.