Immunohistochemical analysis of ABCG2 expression identified two distinct positive populations in the prostate: an endothelial population and a nonendothelial basal epithelial population. Identification of nonendothelial expression was accomplished by comparing serial sections stained with the basal cell marker CD138 to ABCG2 staining patterns. Nonendothelial ABCG2 cells were identified as CD138+/ABCG2+. Nonendothelial staining for ABCG2 was observed in small clusters of cells in the basal epithelium of a few prostate glands. Endothelial CD138-/ABCG2+ staining was observed in capillaries in the stroma, or at periglandular location.
Using a protocol established for mouse bone marrow SP analysis, we isolated an SP in normal human prostate tissue obtained from radical prostatectomies. The forward and side scatter plot revealed a smaller and less granular population of cells corresponding to the SP defined by the Hoechst RED/BLUE profile. Transcriptome analysis of the prostate SP was compared to previously sorted prostate cell transcriptomes [
30]. In contrast to the CD antibody-sorted cell types, far fewer genes were detected in SP sorted cells. RNA integrity is critical for determining gene expression, and suboptimal RNA quality in the SP sorted cells was the probable cause of this decrease in genes detected. Hoechst 33342 is a DNA intercalating agent that results in significant cellular toxicity [
36], and may have caused RNA degradation in cells isolated in this fashion. Several studies have addressed the potential toxicity of Hoechst 33342 [
36,
37], however, very little research has looked at the effects of the SP technique on the quality of RNA isolated from these cells. To mitigate the effects of Hoechst toxicity on RNA degradation, the monoclonal antibody, clone 5D3, was used to generate a transcriptome for the ABCG2
+ prostate epithelial cells. MACS was chosen as the separation method because the procedure is much faster than FACS (minutes
vs. hours).
Transcriptome analysis of the SP and ABCG2
+ cells from the prostate included the endothelial (CD31
+) population [
30] since these cells also express ABCG2. As expected, all three populations (prostate SP, ABCG2
+ and CD31
+) expressed higher levels of ABCG2 than the other prostate cell types: luminal, stromal, basal, and epithelial (data not shown). Additionally, all three populations contained other ABC transporters including ABCE1, ABCB1, ABCG1 and ABCE1 [see Additional file
1]. The ABCG2
+ (5D3) transcriptome dataset was found to contain many genes present in endothelial cells, although it did contain 232 unique probesets with a raw fluorescence signal >100 [see Additional file
2]. The sorting scheme used to isolate ABCG2
+ cells by presorting with CD31-PE might have been inefficient in removing all CD31
+ cells. However, MACS sorting of ABCG2
+ cells did result in a more complete transcriptome than the SP sorting method [see Additional file
3], as evidenced by the nearly 5-fold increase in number of genes identified. RT-PCR analysis revealed that SP cells expressed higher levels of the putative stem cell markers NES, TERT, and BMI-1, in addition to ABCG2 compared to non-SP cells. In the same analysis, other markers, such as the endothelial marker CD31 [
29] and the prostate stem cell marker CD133 [
6] were expressed at similar levels by both SP and non-SP cells. CD31
+ cells also expressed NES, BMI-1, TERT and CD133. qPCR analysis of ABCG2+ cells verified that the ABCG2
+ sorting resulted in enrichment for these genes as compared to CD31
+ cells. Based on these transcriptome and RT-PCR analyses, the distinction between the prostate endothelial cell and the putative prostate stem cell isolated based on SP expression is not entirely clear. The ABCG2
+ population shares many endothelial genes as well, however, based on transcriptome and qPCR analyses, it seems to be a more distinct population than the SP cells. Because of their expression of ABCG2, endothelial cells are a major potential contaminant of stem cells isolated by methods targeting this marker. Alternative sorting schemes to isolate purer ABCG2
+/CD31
- populations are currently being explored in our lab.
Our study was not designed to provide a functional assay for these putative stem cells. The sorting methods utilized resulted in very small numbers of cells that were insufficient for
in vitro studies. Additionally, the fact that these cells express ABCG2, which is hypothesized to efficiently pump out differentiation inducing molecules [
24], may render them non-responsive to differentiation signaling. This transcriptome data has, however, provided a wealth of targets for further studies aimed at identifying the mechanisms that govern the process of self-renewal and development of prostate stem cells.