Therapeutic Activity of Anti-AXL Antibody against Triple-Negative Breast Cancer Patient-Derived Xenografts and Metastasis Free

Breast cancer is the most common female malignancy in western countries. Triple-negative breast cancers (TNBC) account for 15% to 20% of all breast cancer cases and are defined by the absence of estrogen receptors (ER), progesterone receptors (PR), and HER2 expression. TNBC prognosis is poor and chemotherapy remains the sole therapeutic option (1). No targeted therapy is available for TNBC due to the lack of identified molecular targets in most of these cancers. Moreover, TNBC is associated with a significant higher probability and faster occurrence of relapse compared with other breast cancer subgroups. Epithelial-to-mesenchymal transition (EMT) is believed to be a main contributor to the invasion process (2, 3). During EMT, cells lose some of their epithelial characteristics, typically E-cadherin downregulation, and acquire mesenchymal phenotypic marks, characterized by upregulation of vimentin, N-cadherin, and fibronectin. Transcriptional regulators activated in the EMT program include TWIST1, ZEB1, and SNAIL transcription factors (SNAI1/SNAI2). These regulators are expressed in many tumors and have an important role in breast cancer metastasis formation (4).

Several studies reported that AXL, a member of the TAM (TYRO3, AXL, MER) receptor tyrosine kinase family, has a major role in the control of tumor cell migration and invasion (5–8). AXL was first identified as a transforming gene in cells from patients with chronic myelogenous leukemia (9). Following binding to its ligand, growth arrest–specific 6 (GAS6) leads to TAM receptors dimerization and, consequently, to the activation of various intracellular pathways that regulate many biological processes, such as cell invasion and proliferation, angiogenesis, and drug resistance (10).

High AXL expression has been reported in many cancers and is associated with tumor progression and shorter overall survival (10, 11). In breast cancer, Gjerdrum and colleagues found that AXL expression is increased in metastatic lesions compared with matched primary tumors suggesting a crucial role in breast cancer cell invasiveness and metastatic potential (12). However, it is unclear whether AXL directly induces EMT by regulating EMT effectors or whether it is a target of these EMT regulators. Indeed, AXL downregulation in breast cancer using siRNAs inhibits AKT phosphorylation and decreases cell invasion, motility, and metastasis. On the other hand, ectopic expression of AXL in human mammary epithelial cells leads to a downregulation of E-cadherin expression and increased expression of mesenchymal markers (13). Along similar lines, transfection of SLUG and SNAIL genes in MCF10A breast cancer cells induces the acquisition of mesenchymal markers and loss of epithelial characteristics together with increased AXL expression (12).

In this study, we examined the potential therapeutic effect of a recently developed anti-AXL mAb (20G7-D9; ref. 14) in TNBC cell xenografts and in basal-like patient-derived xenografts (PDX). We observed a dual role of our antibody consisting of inhibiting tumor growth and AXL/GAS6-dependent migration/invasion of cancer cells. In fact, we demonstrated, in vitro, that stimulation of several TNBC cell lines with recombinant human GAS6 (rhGAS6) induced the expression of EMT transcription factors, mesenchymal markers, and phosphorylation of FRA-1, a key transcription factor in cancer cell invasion, migration, and plasticity (15). Treatment with 20G7-D9 resulted in the degradation of AXL receptor, avoiding any AXL/GAS6-dependent signaling to occur and, in vivo, reducing spreading of MDA-MB-231-Luc to bones after intracardiac injection. In conclusion, in AXL-positive TNBC tumors, AXL and GAS6, have a critical role in the maintenance of the EMT phenotype. Therefore, AXL should be considered as a therapeutic target of choice for tumor growth and metastasis inhibition.

The MDA-MB-231, MDA-MB-435, MDA-MB-436, MDA-MB-453, MDA-MB-468, BT549, BT20, HCC38, Hs578T, HBL-100 breast cancer cell lines were obtained from ATCC. Cells were cultured for at most 3 months and then cultures were renewed by thawing a new vial from our master cell bank. Authentication by typical morphology observation and mycoplasma testing using the MycoAlert Mycoplasma Detection Kit (Lonza) were routinely performed. All cell lines were cultured following the ATCC recommendations. Control (sh-Luc) and anti-AXL shRNAs were previously described (14). Luciferase-transfected MDA-MB-231 cells (MDA-MB-231-Luc) were previously described (16). RhGAS6 and the goat anti-human GAS6 antibody were purchased from R&D Systems. The anti-GAPDH antibody was from Millipore. All other antibodies used in Western blotting were from Cell Signaling Technology. The anti-FRA1 siRNA corresponds to the 124–144 coding region relative to the first nucleotide of the start codon (15). The specific anti-human AXL murine IgG1 antibody 20G7-D9 that induces AXL internalization and degradation was previously described (14).

Western blotting was performed using protein lysates from cell lines or PDXs, as previously described (17). Western blot analysis was performed using the G:BOX iChemi imaging system (Syngene). Protein expression was quantitated with the ImageJ software and the Student t test was used to evaluate differences.

Cancer cells were serum-starved in the presence or absence of 50 μg/mL 20G7-D9 overnight. This incubation time was chosen as we previously showed a massive degradation of AXL at this time point (14). After trypsin digestion, cells were counted and migration assays were performed by seeding 3 × 10⁴ cells resuspended in serum-free DMEM in 8-μm pore size BD-Falcon 24 Fluoroblock Transwell inserts. For GAS6-dependent migration, 200 ng/mL rhGAS6 was added to the bottom chamber containing the migration medium (DMEM/5% serum). Invasion assays were carried out using Matrigel Matrix Growth Factor Reduced (BD Biosciences) and 5 × 10⁴ cells. After 4 hours of migration and 24 hours of invasion, cells in the bottom chamber were stained with 4 μg/mL calcein and then counted under an inverted fluorescence microscope.

All in vivo experiments were performed in compliance with the French regulations and ethical guidelines for experimental animal studies in an accredited establishment (agreement no. C34-172-27). MDA-MB-231 or MDA-MB-436 cells (5 × 10⁶) were injected subcutaneously in 6-week-old female athymic mice (Harlan). When tumors reached a minimum volume of 100 mm³, tumor-bearing mice were randomized in different treatment groups (n = 6) and treated with 200 μg of 20G7-D9 or saline by intraperitoneal injection twice a week for four consecutive weeks. Tumors were measured using a caliper and volume was calculated using the formula V = (tumor length × tumor width × tumor depth)/2, until the tumor volume reached 2,000 mm³. For metastasis studies, 5 × 10⁵ MDA-MB-231-Luc cells were injected intracardially in anesthetized BALB/c nude mice. Intraperitoneal treatments with 200 μg 20G7-D9/injection or saline were done twice a week for four consecutive weeks. Bone metastases were quantified by in vivo bioluminescence imaging at day 25, as described previously (16). For PDX models, approximately 60 mm³ of B3804 and B3467 tumors were transplanted in the interscapular fat pads of Swiss nude female mice (Charles River Laboratories) and treated with 200 μg 20G7-D9 twice weekly or saline, when tumor volume reached a minimum volume of 100 mm³. Tumor volumes were measured as described above.

Expression profiling of PDXs was performed using Affymetrix U133plus2. Raw data were normalized using robust multi-array average (RMA). Data were aggregated around HUGO gene symbols with max argument (18). The CIT classification developed by our group was applied using the R script referred in the publication (19). The dataset of 1,809 breast cancer samples was downloaded from http://kmplot.com/analysis/index.php?p=download. The CIT classification R script was run on the 1,809 breast cancer samples. The AXL expression level was displayed for each molecular subtypes (Supplementary Fig. S2). The 254 samples identified as Basal-like were extracted. The 100 genes most correlated with AXL expression in the 254 basal-like tumors were obtained using the gene neighbor command of the GenePattern suite (Broad Institute, Boston) with standard settings (including Pearson correlation). Annotations were made with the GSEA software (Broad Institute, Boston), first with hallmark gene sets (Fig. 3) and then with the MSigDB collections.

A linear mixed regression model was used to determine the relationship between tumor growth and number of days after implantation. The fixed part of the model included variables corresponding to the number of days after implantation and the different treatment groups. Interaction terms were built into the model. Random intercepts and random slopes were included to take into account the time effect. The model coefficients were estimated by maximum likelihood and considered significant at the 0.05 level. Statistical analysis was carried out using the STATA 10.0 software.

Before testing the effect of the 20G7-D9 anti-AXL mAbs in TNBC xenografts, we assessed AXL surface expression by flow cytometry (Fig. 1A) and TAM receptors expression by Western blotting (Supplementary Fig. S1) in 10 TNBC cell lines and found that AXL was significantly expressed in 7. These results are consistent with previous studies reporting that AXL is often upregulated in basal B and occasionally in basal A TNBC (20, 21). Moreover, the cell lines in which AXL was most overexpressed (MDA-MB-231, BT549, HBL-100, BT20, and HS578T) are considered as highly invasive (22). Interestingly, we analyzed AXL mRNA expression in a wide range of breast cancer tumors and we did not observe a significant upregulation of this receptor in basal-like breast tumors in comparison with the other breast cancer subtypes (Supplementary Fig. S2).

Treatment of mice harboring established MDA-MB-436 (basal B, moderate AXL expression) and MDA-MB-231 (basal B, high AXL expression) xenografts with 20G7-D9 for 4 weeks led to a statistically significant growth inhibition in both models compared with saline-treated animals (P = 0.015 and 0.047, respectively). Specifically, the mean tumor volume in 20G7-D9–treated mice was reduced by 62.5% at day 53 in the MDA-MB-436 model, and by 57% at day 68 in the MDA-MB-231 model (Fig. 1B).

With the aim of using a more relevant preclinical model, we also included 10 TNBC PDXs generated in our institute (18). Microarray analysis showed that AXL mRNA was expressed in six PDXs (>50 arbitrary units). For instance, the PDX derived from the B3804 tumor showed a very high expression level of AXL (Fig. 2A). The other four showed minimal or no AXL expression. AXL protein expression, assessed by Western blotting, was well correlated with the mRNA values (Fig. 2B). As AXL (protein and mRNA) was not detected in the B3467 tumor, this PDX was used as negative control for further experiments. Moreover, the four PDXs with the highest AXL protein content (Fig. 2B) also showed the fastest growth rate (average passage duration for the three first passages = 52.3 days) compared with the other PDXs (151.1 days, t test; P = 0.0003; Fig. 2C).

We then investigated the effect of 20G7-D9 on the growth of the AXL-positive B3804 and the AXL-negative B3467 PDXs. At the end of the treatment, B3804 growth was significantly reduced (by 51%) in 20G7-D9–treated mice compared with the untreated group (saline; P < 0.01; Fig. 2D). Conversely, growth of the AXL-negative B3467 PDX was not affected by antibody treatment (Fig. 2E).

As AXL expression has been linked to EMT and metastasis formation (13), we next assessed whether 20G7-D9 could inhibit in vivo migration and invasion of cancer cells. To this aim, one day after intracardiac injection of MDA-MB-231-Luc cells, mice were treated with 20G7-D9 or saline (control) for 4 weeks (biweekly injection) and at day 25, the development of MDA-MB-231-Luc metastases in bone was quantified using a luciferin/luciferase assay. The bioluminescence intensity was significantly lower in the group of mice treated with 20G7-D9 than in untreated controls (Fig. 2F; P = 0.0038). This result was confirmed by bioluminescence imaging performed at the end of the 4 weeks of treatment (Fig. 2G).

The finding that treatment with 20G7-D9 reduces cancer cell invasion to bone in mice with TNBC cell xenografts was in accordance with the AXL gene being a major actor in EMT induction. Accordingly, Western blot analysis of the expression of EMT markers involved in breast cancer metastasis formation in TNBC PDXs showed that all AXL-positive PDXs expressed FRA1, SNAIL, and TWIST1, whereas SLUG was detected only in three of them (Fig. 2B). Generally, all AXL-positive PDXs expressed molecules involved in EMT, with the exception of GAS6, MMP1, and VIMENTIN that were not correlated with AXL expression. Moreover, microarray analysis of the 100 genes most correlated with AXL expression in 254 basal-like breast cancers (Supplementary Fig. S3) using the MSigDB hallmark gene signatures indicated a strong enrichment in genes that define the EMT and TGFB1 pathways (Fig. 3A). Particularly, 8 of the 12 genes most correlated with AXL expression, were annotated as EMT markers and included TGFB1 and VEGFC (Fig. 3B). Further annotation analysis with the MSigDB collections indicated a strong enrichment in signal transduction and cell proliferation. Membrane components, including the integrin pathway and focal adhesion processes, were also overrepresented. Significant overlaps were also found with gene signatures present in adult stem cells, early embryonic stem cells, or epithelial cell lines submitted to EMT-inducing factor TGFB1 and mesenchymal breast cancer cell lines. These enrichments reveal the influence of AXL in human tumors through integrin-mediated signaling impacting cell proliferation and EMT in accordance with our in vitro experiments (Supplementary Table S1). These findings in basal-like breast cancer samples and PDXs suggest that high AXL expression is associated with acquisition of mesenchymal features. The extent of this effect might depend on the cell context and tumor environment.

To further understand how 20G7-D9 inhibits metastasis formation, we then performed Boyden chamber assays using the AXL-positive MDA-MB-231 and HBL-100 cell lines (basal B TNBC). Preincubation with 20G7-D9 inhibited migration and invasion by more than 50% in both lines compared with untreated cells (Fig. 4A and B). Conversely, addition of rhGAS6 in the bottom chamber increased significantly the migration and invasion of both cell lines compared with untreated cells (Fig. 4A and B). This effect was abolished by preincubation with the anti-AXL antibody. We then assessed whether 20G7-D9 could inhibit degradation of the extracellular matrix by breast cancer cells using in situ zymography. Preincubation of MDA-MB-231 cells with 20G7-D9 significantly reduced the number of invadopodia and the areas of focal gelatin degradation compared with untreated cells (Supplementary Fig. S4A and S4B). Collectively, these data suggest that GAS6 and AXL are involved in TNBC cell migration and invasion.

To further investigate the role of GAS6/AXL signaling in EMT and metastasis formation, we incubated MDA-MB-231, HBL-100 (two cell lines with high AXL expression), and MDA-MB-436 (moderate AXL expression) cells with rhGAS6 and then assessed AXL and FRA-1 phosphorylation and the expression of EMT effectors by Western blotting. GAS6 induced phosphorylation of AXL and FRA-1 (at S265) and increased AXL expression in all three cell lines (Fig. 5A). Moreover, expression of SNAIL, SLUG, TWIST1, and ZEB-1/2 was also increased in the three TNBC cell lines upon addition of GAS6 (Fig. 5A and B, quantification in Supplementary Fig. S5). These effects were abolished when AXL was knocked down by shRNA in MDA-MB-231 cells. Similarly, FRA1 silencing in MDA-MB-231 cells inhibited ZEB1 and SNAIL upregulation following stimulation with GAS6 (Fig. 5C). This is in agreement with the finding that FRA-1 can induce EMT-related genes, such as SLUG and ZEB1/2 (23, 24), and suggests that some of the effects of activated AXL could result from increased expression of phosphorylated FRA-1.

Finally, incubation with GAS6 also induced expression of the mesenchymal markers vimentin, MMP1 and MMP9 in HBL100 and MDA-MB-436 cells (Fig. 5B). In MDA-MB-231 cells, induction was less clear-cut due to the high basal expression of these markers. Nevertheless, AXL silencing reduced the expression level of the three EMT markers (Fig. 5B). Moreover, the level of induction of these EMT markers varied according to the TNBC cell line and could be explained by the involvement of other, still unidentified factors in AXL/GAS6 signaling. Globally, these results confirm an important role of GAS6 and AXL in the expression of EMT markers involved in breast cancer invasion.

We then asked whether 20G7-D9 could influence EMT marker expression and migration of MDA-MB-231 cells. Incubation with 20G7-D9 induced downregulation of AXL (14) and consequently its expression could not be induced by GAS6 (Fig. 6A and D). Similarly, GAS6-induced AXL phosphorylation was reduced compared with untreated cells.

A significant inhibition of the GAS6-induced upregulation of the EMT inducers SNAIL, SLUG, TWIST, and ZEB1/2 as well as of FRA-1 was observed in MDA-MB-231 cells preincubated with 20G7-D9 (Fig. 6B and D) compared with untreated cells. GAS6-induced FRA-1 phosphorylation was also strongly inhibited upon AXL downregulation by 20G7-D9. Finally, the anti-AXL antibody treatment also reduced GAS6 effect on the expression of the mesenchymal markers vimentin, MMP1, and MMP9 (Fig. 6C and D). In BT20 cell line, GAS6 treatment induced a strong downregulation of E-cadherin that is abrogated by pre-treatment with 20G7-D9 anti-AXL antibody (Supplementary Fig. S6).

Among the large number of drug approved for breast cancer, the use of targeted therapies [hormonal, tyrosine kinase inhibitors (TKI), or antibodies] highlighted the existence of an aggressive breast cancer subtype, the TNBC, characterized by frequent and rapid relapse. Although this subtype is sensitive to chemotherapy, the risk of relapse and metastasis formation is high and the development of novel therapies to suppress cancer metastasis is needed. In this study, we show that, in vitro, AXL and its ligand GAS6 enhance TNBC cell migration and invasiveness. Furthermore, we also demonstrate that after activation by GAS6, AXL promotes EMT by upregulating the transcription factors FRA1, SLUG, SNAIL, ZEB1/2, and TWIST1. AXL also increases cell motility by inducing expression of MMP1 and MMP9, thus promoting cell migration and invasion. Inhibition of AXL by shRNA or with the anti-AXL mAb 20G7-D9 suppresses both cancer cell migration and invasion, even following GAS6 addition to the medium.

In vivo treatment with 20G7-D9 reduces tumor growth in TNBC xenograft models and also in AXL-positive basal-like breast cancer PDXs. A limitation of cancer cell line xenograft models is that the genetic and physiologic backgrounds of these cell lines strongly diverge from those of primary tumors (25). Moreover, tumor heterogeneity is often lost after cell culture, whereas it is an important characteristic of breast cancer and can affect the tumor response to treatment. PDX are today, with genetically engineered cancer mouse models, a more relevant model because they are based on the graft of fresh primary tumor fragments directly from the patient to immunodeficient mice. PDX maintain the genetic and histologic features of the original tumor (18, 26, 27). For instance, a good prediction of patient treatment efficacy using PDXs has been reported (28). Nevertheless, it is important to note that, even if PDXs are today recognized as a robust in vivo model for preclinical assays, the microenvironment remains of murine origin. This context might have an effect on AXL expression and role that only clinical trial, targeting the receptor with an antibody or a TKI, would be able to answer. Today, various AXL-specific TKIs (5, 29, 30) and mAbs (14, 31, 32) have been tested in preclinical models and only one AXL inhibitor, BGB324 (R428), is evaluated in clinical trials for acute myeloid leukemia (33). Phase I studies demonstrated that BGB324 could be safely administered for prolonged periods at doses that inhibit AXL activation and exhibit antileukemic activity.

In the third in vivo experiment, we demonstrated the importance of AXL in cancer cell migration and invasion. Indeed, treatment with 20G7-D9 after intracardiac injection of MDA-MB-231-Luc cells significantly reduced the number of bone metastases. Therefore, in the context of AXL-expressing TNBC, AXL seems a valuable therapeutic target because it drives both tumor growth and cancer cell migration/invasion.

In cancer, EMT involves complex cell signaling cascades that result in the alteration of cell–cell and cell–extracellular matrix interactions, leading to the acquisition by epithelial cells of a mesenchymal phenotype and ultimately to metastasis formation (34). Among the EMT cellular pathways, receptor tyrosine kinases can be inducers (for instance, c-MET), or can be activated (for instance, PDGFRα) following induction of the expression of different EMT transcription factors (35, 36). In breast cancer, AXL was known as an EMT-induced regulator of cancer metastasis, under the control of SLUG (13). Although AXL is required to maintain the invasiveness of mesenchymal-like cells, EMT starting point is the overexpression of intracellular proteins, such as SLUG or RAS (12, 37). More recently, Asiedu and colleagues described AXL as an inducer of EMT using the TKI MP470 (13). AXL inhibition led to the downregulation of EMT markers, such as SNAIL and N-cadherin. However, these studies did not assess the role of GAS6, and MP470 can target different tyrosine kinases that have been already defined as EMT inducers (for instance, c-KIT; ref. 38). Here, we show that AXL activation by GAS6 is necessary to promote strong EMT signaling through the involvement of many EMT transcription factors. Moreover, AXL inhibition by shRNA or with the specific antibody 20G7-D9 blocks GAS6/AXL-dependent EMT. Thus, AXL may play a dual role in TNBC as inducer and target of EMT signaling during metastasis formation. This GAS6-induced signaling, as well as the basic expression of EMT markers, vary in the different TNBC cell lines studied and could imply other intracellular factors that remain to be identified.

Furthermore, we show the involvement of the transcription factor FRA-1, an AP-1 family member, in GAS6/AXL signaling. Several studies suggest that FRA-1 can regulate cancer cell motility and invasion. Direct or indirect transcriptional FRA-1 targets include proteins that promote extracellular matrix degradation (MMP1, MMP9; ref. 15), EMT transcription factors (SLUG and ZEB1/2; refs. 24, 39), and proteins involved in cell adhesion or migratory signaling (CD44, VEGF, c-MET; ref. 40). In bladder cancer, FRA-1 promotes AXL transcription that mediates FRA-1 effect on cancer cell motility (41). Our findings that in TNBC cell lines AXL activation by GAS6 increases FRA-1 expression and FRA1 knockdown abolishes SNAIL and ZEB1 upregulation by GAS6/AXL signaling bring new insights into the cross-talk between these proteins in EMT signaling.

Although TNBC cell lines are frequently AXL-positives, we observed that AXL expression in patients was not specific to this breast cancer subtype, confirming previous studies (22). It would be thus interesting to extend the characterization of AXL, as well as the therapeutic use of our antibody, in other breast cancer subtypes, like luminal models, to know whether the receptor has the same behavior and implication in EMT signaling.

Several of the basal-like PDXs we established express elevated levels of AXL and interestingly those with the highest AXL expression level grow faster in vivo. Moreover, analysis of the genes correlated with AXL upregulation in basal-like breast tumors indicated that most of these genes are involved in cancer cell invasion and EMT, TGFβ signaling, mesenchymal markers, stem cells, cytoskeleton, and ECM remodeling. Of note, the transcription factor FRA-1 induces EMT through TGFβ modulation (42). AXL is also overexpressed in breast cancer stem cells (13) and EMT features are often associated with stemness (43).

In conclusion, our results highlight the expression and role of AXL and its ligand GAS6 in TNBC cell invasion. Our results suggest that AXL is a modulator, and not only a target, of EMT signaling. The use of anti-AXL antibodies, which degrade the receptor, is an interesting approach to inhibit metastatic process in such breast cancers.

W. Leconet, C. Larbouret, A. Pèlegrin, and B. Robert are listed as coinventors on a patent on the development of a murine antibody (20G7-D9) against the human receptor tyrosine kinase AXL for cancer therapy, which is owned by the French Research Agency of Health (INSERM), University of Montpellier, and Regional Clinical Cancer Center (ICM), and licensed to Oribase-Pharma. No potential conflicts of interest were disclosed by the other authors.

Conception and design: W. Leconet, D. Chalbos, A. Pèlegrin, C. Larbouret, B. Robert

Development of methodology: W. Leconet, A. Sirvent, M. Busson, N. Radosevic-Robin

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): W. Leconet, M. Chentouf, S. du Manoir, I. Aït-Arsa, M. Busson, N. Radosevic-Robin, C. Theillet, D. Chalbos, J.-M. Pasquet

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W. Leconet, S. du Manoir, M. Busson, M. Jarlier, B. Robert

Writing, review, and/or revision of the manuscript: W. Leconet, S. du Manoir, C. Theillet, D. Chalbos, A. Pèlegrin, C. Larbouret, B. Robert

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W. Leconet, C. Chevalier, B. Robert

Study supervision: W. Leconet, B. Robert

The authors wish to thank Dr. Rui Bras-Gonçalves for his expert help with the PDXs. We thank all the staff of the IRCM animal facility, the RHEM, and the Small Animal Imaging Platform of Montpellier (IPAM) facilities for IHC and mice imaging, respectively. We thank also to Dr. Pierre Martineau, Dr. Antonio Maraver, and Dr. Claude Sardet for their input concerning the manuscript. We thank Béatrice Orsetti for supplying PDX samples for Western blot experiments.

This publication has been funded with support from the French National Research Agency under the program “Investissements d'avenir” Grant Agreement LabEx MAbImprove: ANR-10-LABX-53. This work was supported by the INSERM; Oribase Pharma. This work benefited from the support of INCa grants MOPRECLI and “Role of cancer stem cells during metastatic progression of breast cancer.”

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.