The stem cell inhibitor salinomycin decreases colony formation potential and tumor‐initiating population in docetaxel‐sensitive and docetaxel‐resistant prostate cancer cells

Abstract Background Prostate cancer (PCa) is one of the most frequently diagnosed tumors in men. In general, therapies for localized PCa are curative. However, treatment of advanced PCa is considered palliative since development of therapy resistance occurs rapidly. It has been shown that tumor‐initiating cells are likely involved in therapy resistance. They are not eliminated by conventional therapies and thereby lead to tumor progression and relapse. The aim of this study was to evaluate the effects of the known stem cell inhibitor salinomycin on this critical subpopulation of cells. Methods Expression of the cell surface markers CD24 and CD44 was assessed by immunofluorescence and fluorescence‐activated cell sorting. Colony formation efficiency and classification of colony types with varying tumor‐initiating potential (holoclones, meroclones, and paraclones) were analyzed in an automated way by the newly developed CATCH‐colonies software in the absence or presence of salinomycin. Results Automated high‐resolution colony formation analysis consistently identified the various colony types in a broad range of PCa cell lines. Serial clonogenic assays confirmed that holoclones show the highest colony formation potential and maintain their tumor‐initiating capacity over multiple rounds. Furthermore, holoclones showed high expression of CD44, while CD24 was not expressed in these clones, thus representing the well‐described tumor‐initiating CD24−/CD44high population. Salinomycin decreased the CD24−/CD44high population in both docetaxel‐sensitive PC3 and docetaxel‐resistant (DR) PC3‐DR. Moreover, treatment of PC3, DU145, PC3‐DR, and DU145‐DR with salinomycin led to a significant reduction in the colony formation potential by targeting the colonies with high tumor‐initiating potential. Conclusions Taken together, we demonstrated that salinomycin specifically targets the tumor‐initiating cell population in docetaxel‐sensitive and docetaxel‐resistant PCa cells and may represent a potential therapeutic approach for the treatment of advanced PCa.


| INTRODUCTION
Prostate cancer (PCa) is one of the most frequently diagnosed malignant tumors in men. Although therapy of localized PCa is curative, treatment of advanced PCa is considered palliative. Therapies for patients with biochemical and clinical recurrence predominantly target the androgen receptor (AR). However, these treatments inevitably lead to the development of castration-resistant PCa within a few years. A possible explanation for development of resistance to androgendirected therapies is the existence of cancer stem cells, which were described in detail by Maitland et al. 1 They postulate that there is a hierarchy within tumors and that the bulk population of tumor cells is derived from tumor-initiating cells, which represent a small selfrenewing subpopulation. 2 They may belong to the basal compartment of the prostate, 1 show no AR expression and thereby are not targeted by conventional therapies. 3 The issue of the origin of PCa is still open for discussion. Other researchers have described the role of prostate luminal progenitor cells in tumorigenesis. 4 However, there is still much discussion on how these aggressive cells can be identified and consequently targeted. Barrandon  showed that holoclones display the highest tumorigenic potential when inoculated into mice, in contrast to paraclones.
The aim of this study was to analyze the effects of the stem cell inhibitor salinomycin, which has been described by Dewangan et al. 12 In particular, we determined the impact of salinomycin on the tumorinitiating CD24 − /CD44 high population. 13 Furthermore, we evaluated the effects of salinomycin on colony formation efficiency and distribution of colony types in docetaxel-sensitive and docetaxel-resistant cells using automated high-resolution colony formation analysis. Salinomycin has been used in docetaxel-resistant cells because those cells are known to express stem-like properties 14,15 and the combination of salinomycin and docetaxel was earlier proposed to be a promising strategy to target both gastric cancer cells and cancer stem cells. 16 2 | MATERIALS AND METHODS

| Statistical analysis
Statistical analysis was performed using GraphPad Prism 5 (Graph-Pad Software Inc., San Diego, CA). Differences between control and treatment groups were analyzed using the Student t test. P < .05 was considered statistically significant and encoded as follows: *P < .05; **P < .01. All experiments have been performed in at least three biological replicates.

| Identification, characterization, and automated analysis of colony types in PCa cells
To investigate the three previously described colony types, 5 several PCa cell lines were seeded at low density and grown for 10 to 14 days. In general, the three colony types were found in all cell lines tested with the exception of PC3. Figure 1 shows representative images of the colony types with contrasting morphologies. Paraclones have a quite irregular structure with loosely packed cells that show the highest grade of differentiation. Holoclones, on the other hand, are tightly packed and very compact. Meroclones are semisolid, but they do not have the same dense structure as holoclones.
In this study, the classification of colony types was performed in an automated way using the software CATCH-colonies. In all cell lines, paraclones form the largest segment (from 56% in PC3 to 86% in LAPC4, Figure 2A), followed by meroclones (from 10% in LAPC4 to 43% in PC3). Holoclones form the smallest part (from 0% in PC3 to 15% in DU145), which is in agreement with other publications. 7 Figure S1A). Subsequently, we performed serial clonogenic assays to further confirm the correct annotation of the individual colony types. Paraclones showed the lowest colony formation efficiency in PC3 and DU145 cells and lost their proliferative potential after a few passages ( Figure 2C). In contrast, meroclones and holoclones could be passaged over numerous rounds, which confirms their self-renewing potential. Of note, in PC3 cells did not give rise to classical holoclones with a dense inner core; however, meroclones in PC3 maintained a very high colony formation efficiency similar to holoclones in DU145 cells ( Figure 2C).
Subcultivation of single colonies predominantly led to the formation of daughter-colonies of their respective type ( Figure S1B). Most importantly, only holoclones were able to give rise to all three types of colonies which has also been observed by others. 7

| Salinomycin treatment reduces tumorinitiating CD24 − /CD44 high population
To validate the connection between identified colony types and the  (Figures 3B and 3C). Many patients continue to progress after a short period of time and therapy resistance emerges quickly. 18 Hence, the treatment of advanced PCa remains a major issue and there is an urgent need to identify new therapeutic options to overcome therapy resistance. Common treatments for advanced stages of PCa include androgen deprivation therapy, inhibitors of androgen synthesis and anti-androgens such as enzalutamide and abiraterone. All these therapies target the AR in highly proliferative cells. Conventional therapies are inefficient in eliminating stem cells, which are AR-negative 3 and only show low proliferation and apoptosis rates. 19 Therefore, it is important to find novel treatment options that eliminate the small population of tumor-initiating cells that represent the top of the hierarchy in the bulk of PCa cells.

| Salinomycin suppresses the formation of colonies with high tumor-initiating potential
There is still much discussion on how tumor-initiating PCa cells can be identified and many approaches already exist. In this study, the classification of colony types was performed automatically by the CATCH-colonies software, which eliminates subjective F I G U R E 2 Automated classification of colony types by the software CATCH-colonies. A, Quantification of the three colony types (red, paraclones; blue, meroclones; green, holoclones) in PC3, DU145, LNCaP, and LAPC4. B, Clustering of different colony types by principal component analysis (PCA) after analysis by the CATCH-colonies software. C, Serial clonogenic assays of colony types in PC3 and DU145 (n = 1, the experiment was performed to ensure that identification of the colony types is in accordance with the work of others 7 ) [Color figure can be viewed at wileyonlinelibrary.com] characterization and leads to reproducible results. The automated classification was confirmed by quantitative real-time PCR analysis of the stem cell-related genes Nanog, ALDH1A3, and OCT4 and serial clonogenic assays. Moreover, we demonstrated by immunofluorescence staining that holoclones represent the tumor-initiating CD24 − /CD44 high population that has previously been described by Al-Hajj et al. 13 The therapeutic compound salinomycin is an antibacterial drug that is naturally produced by Streptomyces albus and has previously been used as coccidiostat in animals. 20 The mechanism of action is still not fully elucidated and numerous pathways have been described to be targeted by salinomycin. 12 Salinomycin has been reported to exert anticancer effects in several tumor entities including PCa. [21][22][23][24] In detail, it has been shown that gastric cancer stem cells, which were characterized by enhanced Wnt/β-catenin signaling, are targeted by salinomycin supporting its activity against tumor-initiating cells. 25 Moreover, it was demonstrated that salinomycin also decreased the CD24 − /CD44 + stem-like population in breast cancer cells 26   F I G U R E 4 Salinomycin decreases colony formation efficiency by targeting tumor-initiating clones. A, Colony formation efficiency of PC3, PC3-DR, DU145, and DU145-DR treated with the indicated concentrations of salinomycin was assessed by determining colony numbers after 10 days. Quantification was performed by the CATCH-colonies software. Data represent mean ± SEM (*P < .05; **P < .01; t test). B, Clustering of colony types in PC3 treated with the indicated concentrations of salinomycin (paraclones, red; meroclones, blue). C, Relative number of each colony type in PC3 and PC3-DR upon salinomycin treatment was analyzed by the software CATCH-colonies. Values indicated are mean ± SEM (*P < .05; t test). D, Clustering of colony types in DU145 upon salinomycin treatment (paraclones, red; meroclones, blue; holoclones, green). E, Relative number of each colony type in DU145 and DU145-DR upon salinomycin treatment. Data represent mean ± SEM (*P < .05; **P < .01; t test) [Color figure can be viewed at wileyonlinelibrary.com]