(A) To determine whether the bystander killing effect induced by repertaxin treatment was mediated by FASL, we measured the level of soluble FASL in the medium using an ELISA assay. After 4 days of treatment, a greater than 4-fold increase of soluble FASL was detected in the medium of cells treated with repertaxin compared with untreated controls. (B) We measured the level of FASL mRNA by RT-PCR and confirmed the increase of FASL production after treatment with repertaxin. Similar results were observed after 4 days of treatment with a FAS agonist that activates FAS signaling, with a 5-fold increase of the FASL mRNA compared with control. (C) SUM159 cells were cultured in adherent conditions and treated with repertaxin alone or in combination with anti-FASL. Interestingly, cell growth inhibition induced by repertaxin treatment was partially rescued by addition of anti-FASL. Moreover, cells treated with a FAS agonist displayed cell growth inhibition similar to that of cells treated with repertaxin alone. (D and E) The effect of repertaxin treatment, alone or in combination with anti-FASL, and FAS agonist treatment on the CXCR1⁺ and ALDEFLUOR⁺ population was analyzed. The massive decrease in the CXCR1⁺ and ALDEFLUOR⁺ population induced by repertaxin treatment was not rescued by the anti-FASL, and treatment with FAS agonist produced 10- and 3-fold increases in the percent of the CXCR1⁺ and ALDEFLUOR⁺ populations, respectively. BAAA, BODIPY aminoacetaldehyde; DEAB, diethylaminobenzaldehyde. Error bars represent mean ± SD.

(A–C) For each xenograft, 50,000 cells were injected into the mammary fat pad of mice. (A) Tumor size before and during the course of each indicated treatment (arrow indicates beginning of the treatment). Similar results were observed for each sample (UM2 shown here; see Supplemental Figure 12 for SUM159, MC1, and UM3), with a statistically significant size reduction of the tumor treated with docetaxel alone or in combination with repertaxin compared with the control tumors (P < 0.01). (B and C) Docetaxel-treated tumor showed similar or increase percentage of ALDEFLUOR⁺ cells compared with the control, whereas repertaxin treatment alone or in combination produced a statistically significant decrease in ALDEFLUOR⁺ cells (P < 0.01; B). Serial dilutions of cells obtained from these xenografts were implanted in the mammary fat pad of secondary mice, which received no further treatment. Cells from control and docetaxel-treated tumors formed secondary tumors at all dilutions, whereas only higher numbers of cells obtained from xenografts treated with repertaxin alone or in combination were able to generate tumors (P < 0.01; C). (D) Xenotransplants from each group were collected, and immunohistochemistry staining was done. Phospho-FAK, phospho-AKT, and ALDH1 expression was detected in the control and docetaxel-treated tumors, whereas low or no expression was detected in the tumors treated with repertaxin alone or in combination. Nuclear FOXO3A expression was detected in the cells treated with docetaxel and/or with repertaxin (arrowheads denote positive staining). Scale bar: 100 μm. Error bars represent mean ± SD.

(A) Potential IL-8/CXCR1 cell signaling in CSCs. CXCR1 activation upon IL-8 binding induces FAK phosphorylation. Active FAK phosphorylates AKT and activates the WNT pathway, which regulates stem cell self-renewal and FOXO3A that regulates cell survival. Activation of FAK protects CSCs from a FASL/FAS-mediated bystander effect by inhibiting FADD, a downstream effector of FAS signaling. In the presence of chemotherapy, only the bulk tumor cells are sensitive to the treatment and release a high level of IL-8 and FASL proteins during the apoptotic process. Breast CSCs are stimulated via an IL-8–mediated bystander effect and are resistant to the bystander killing effect mediated by FASL. TCF, T cell factor. (B) Repertaxin treatment blocks IL-8/CXCR1 signaling and inhibits breast CSC self-renewal and survival. When repertaxin treatment is combined with chemotherapy, the CSCs are sensitized to the bystander killing effect mediated by FASL.

(A) Representation of the overlap between the ALDEFLUOR⁺ subpopulation and the CXCR1⁺ (top) or CXCR2⁺ (bottom) subpopulation of SUM159 cells. (B and C) SUM159 cells were cultured in adherent conditions and treated with repertaxin and 2 specific blocking antibodies for CXCR1 or CXCR2. After 3 days, the effect on cell viability and the CSC population was analyzed. A significant reduction of the ALDEFLUOR⁺ population and cell viability was observed after treatment with repertaxin or anti-CXCR1 antibody, but not with anti-CXCR2 antibody. (D) After 4 days of treatment, the number of apoptotic cells was evaluated, and 36% of apoptotic cells (green) were detected in repertaxin-treated cells compared with controls, in which mostly viable cells (blue) were present. Scale bars: 100 μm. (E) To determine whether cell death was mediated via a bystander effect, CXCR1⁺ and CXCR1^– populations were treated with various concentrations of repertaxin. A decrease in cell viability in CXCR1⁺ and unsorted populations were detected, whereas no effect was observed in the CXCR1^– population. (F) Serial dilutions of dialyzed conditioned medium from CXCR1⁺ cells treated for 3 days with repertaxin was used to treat sorted CXCR1⁺, CXCR1^–, or unsorted populations. After 2 days of treatment, a massive decrease in cell viability was observed in both CXCR1^– and unseparated populations, whereas no effect was observed in the CXCR1⁺ population. Error bars represent mean ± SD.

To evaluate the effect of repertaxin treatment on CXCR1 downstream signaling, we used 2 different viral constructs, 1 knocking down PTEN expression via a PTEN-siRNA and the other leading to FAK overexpression (Ad-FAK). (A) Repertaxin treatment led to a decrease in FAK Tyr³⁹⁷ and AKT Ser⁴⁷³ phosphorylation, whereas PTEN deletion and FAK overexpression blocked the effect of repertaxin treatment on FAK and AKT activity. (B) Using immunofluorescence staining on CXCR1⁺ cells, we confirmed that repertaxin treatment caused a disappearance of phospho-FAK (membranous staining in red) and phospho-AKT expression (cytoplasmic staining in red). Immunofluorescence staining with anti-FOXO3A revealed a cytoplasmic location of FOXO3A (red) in the untreated cells, whereas repertaxin treatment induced a relocalization of FOXO3A to the nucleus. In contrast, cells with PTEN deletion or FAK overexpression displayed a high level of phospho-FAK, phospho-AKT, and cytoplasmic FOXO3A expression in both the repertaxin-treated and untreated cells. In all samples, nuclei were counterstained with DAPI (blue). Scale bar: 50 μm. (C and D) The effect of repertaxin on SUM159 PTEN-siRNA and SUM159 Ad-FAK cell viability and on the CSC population was assessed using MTT and ALDEFLUOR assays, respectively. After 3 days of treatment, cells with PTEN deletion or FAK overexpression developed resistance to repertaxin (C). Furthermore, repertaxin treatment did not alter the proportion of ALDEFLUOR⁺ SUM159 PTEN knockdown cells (D). Error bars represent mean ± SD.

(A–C) To evaluate the effect of repertaxin treatment on metastasis formation, we infected HCC1954, SUM159, and MDA-MB-453 breast cancer cell lines with a lentivirus expressing luciferase, and inoculated 250,000 luciferase-infected cells into NOD/SCID mice via intracardial injection. Mice were treated 12 hours after intracardiac injection with either s.c. injection of saline solution or s.c. injection of 15 mg/kg repertaxin twice daily for 28 days. Metastasis formation was monitored using bioluminescence imaging. Quantification of the normalized photon flux, measured at weekly intervals following inoculation, revealed a statistically significant decrease (P < 0.01) in metastasis formation in repertaxin-treated compared with saline controls for mice inoculated with HCC1954 (A) or SUM159 (B) cells. In contrast, repertaxin treatment did not have any effect on metastasis formation for the mice injected with MDA-MB-453 (C) cells. (D) Histologic confirmation, by H&E staining, of metastasis in bone and soft tissue resulting from mice not treated with repertaxin. Scale bar: 100 μm. Error bars represent mean ± SD.