Melanoma chemotherapy leads to the selection of ABCB5-expressing cells

Abstract

Metastatic melanoma is the most aggressive skin cancer. Recently, phenotypically distinct subpopulations of tumor cells were identified. Among them, ABCB5-expressing cells were proposed to display an enhanced tumorigenicity with stem cell-like properties. In addition, ABCB5(+) cells are thought to participate to chemoresistance through a potential efflux function of ABCB5. Nevertheless, the fate of these cells upon drugs that are used in melanoma chemotherapy remains to be clarified. Here we explored the effect of anti-melanoma treatments on the ABCB5-expressing cells. Using a melanoma xenograft model (WM266-4), we observed in vivo that ABCB5-expressing cells are enriched after a temozolomide treatment that induces a significant tumor regression. These results were further confirmed in a preliminary study conducted on clinical samples from patients that received dacarbazine. In vitro, we showed that ABCB5-expressing cells selectively survive when exposed to dacarbazine, the reference treatment of metastatic melanoma, but also to vemurafenib, a new inhibitor of the mutated kinase V600E BRAF and other various chemotherapeutic drugs. Our results show that anti-melanoma chemotherapy might participate to the chemoresistance acquisition by selecting tumor cell subpopulations expressing ABCB5. This is of particular importance in understanding the relapses observed after anti-melanoma treatments and reinforces the interest of ABCB5 and ABCB5-expressing cells as potential therapeutic targets in melanoma.

Figure 1. ABCB5 is expressed on the surface of a subpopulation of melanoma cells.

WM-266-4 cells were surface-labelled with the ABCB5-Ab^Rock antibody and analyzed by flow cytometry. ABCB5⁺ cells (right contour plot) were gated on viable cells (DAPI-negative) according to the isotype control (left contour plot). Inserts (dot plots) display the gating of the positive cells (A). WM-266-4 cells were treated with a siRNA designed to target ABCB5 (si-ABCB5). After 72 h, the cells were analyzed for their ABCB5 mRNA content and ABCB5 surface expression. The left and right histograms show respectively the relative expression of ABCB5 mRNA normalized to the ABCB5 mRNA in cells treated with a control siRNA (si-ctrl), and the percentage of ABCB5⁺ cells among total cells (n = 3). The corresponding contour plots are shown (B). Different melanoma cell lines were analyzed for their ABCB5 surface expression (C) or their ABCB5 mRNA content (D) (n = 3).

Figure 2. ABCB5-expressing cells are enriched in the residual tumors after an anti-melanoma treatment in vivo.

WM-266-4 cells (5×10⁶ cells) were injected subcutaneously in Swiss nude mice. Fourteen days later, mice were treated by repeated i.p. injections of either temozolomide (80 mg/kg) or vehicle following the schedule indicated by the black arrows. The tumoral volumes were monitored and the means of measured volumes respectively for temozolomide-treated and vehicle treated tumors are as follows: 180 mm³ and 175 mm³ at day 14; 264 mm³ and 295 mm³ at day 16; 178 mm³ and 444 mm³ at day 18; 84 mm³ and 699 mm³ at day 21. (A). 24 h after the injections at days 16 and 21 (d17 and d22), tumors were recovered, dissociated and the cell suspensions were searched for the presence of human ABCB5⁺ cells by flow cytometry (B) (medians are represented as black lines). Tumors recovered at day 17 were analyzed by immunohistochemistry for their ABCB5 expression (C).

Figure 3. ABCB5 expression is increased in melanoma tumors obtained from treated patients.

Skin metastases specimens from respectively 8 untreated and 7 treated patients were analyzed by immunohistochemistry for their ABCB5 protein expression. The ABCB5 staining intensity was ranked in four arbitrary classes according to the intensity and the extent of the labelling. Representative staining of two levels of intensity (left panel: isotypic control) (A). Repartition of the specimens in the different classes (B). The two groups of specimens (untreated versus treated) have been compared with the non parametric Kruskall Wallis test (p<0.30).

Figure 4. ABCB5-expressing cells survive upon dacarbazine treatment.

WM-266-4 (A,D,G), G-361 (B,E,H) and SK-MEL-28 cells (C,F,I) were treated at the indicated concentrations of dacarbazine (A–C), vemurafenib (D–F) and doxorubicin (G–I). After 72 h, the total viable cells were numbered using an automated cell viability analyzer. The percentages of viable ABCB5-expressing cells (among viable cells gated on DAPI-negative cells) were analyzed by flow cytometry. The numbers of total cells (white symbols) are reported as percentages of the number of cells in the untreated control sample. The numbers of viable ABCB5⁺ cells (black symbols) were calculated from the total cell numbers and ABCB5⁺ cells ratio, and reported as percentages of the viable ABCB5⁺ cells number in the control sample. ABCB5⁺ cells represent respectively 3%, 3.5% and 5% of the total cells in the WM-266-4, G-361 and SK-MEL-28 cell lines.

Figure 5. Enrichment in ABCB5-expressing cells is associated with protein neo-synthesis and ABCB5 relocation at the cell-surface.

WM-266-4 cells were treated with dacarbazine, doxorubicin or vehicle (NT) for 72 h. Cycloheximide (A–B) or brefeldin A (C–D) were added respectively 24 h and 4 h before the treatment end-point. Cells were labelled for ABCB5 and analyzed by flow cytometry. The means of fluorescence intensity of the ABCB5⁺ cells from three independent experiments were reported in A and C. Fluorescence intensity histograms of representative experiments are shown in the figure S1. The number of ABCB5⁺ cells was reported in B and D as a percentage of the ABCB5⁺ cell number in the vehicle-treated sample.

Figure 6. Quantification of ABCB5- and ABCB1-expressing cells after cytotoxic treatments.

WM-266-4 cells were treated for 72 h with the indicated concentrations of doxorubicin, dacarbazine, vemurafenib, gemcitabine or with vehicle (NT) and ABCB5 expression was analyzed by Western blot. Band intensities were quantified and variations are indicated as fold increases in treated versus untreated samples (A). WM-266-4 cells were treated with various drugs at their EC50 for 72 h. The percentages of positive cells among surviving cells were measured by cell surface labelling and flow cytometry analysis for ABCB5 (B) or ABCB1 (C). The relative mRNA expression of ABCB5, ABCB1, ABCC1, ABCG2 and HMBS as the house-keeping gene was measured by Q-PCR (see also table S1) and the amplified products were run on agarose gel after 29 cycles except for ABCB1 (32 cycles) (D).