*2.5. Three-Dimensional Spheroid Invasion Assay*

Invasion assays using 3D spheroids in the presence or not of dasatinib or PF-573228 (Selleckchem, Houston, TX, USA) were performed as previously described [30] (see Supplementary Materials information for details).

#### *2.6. In Vivo Tumor Growth*

All animal research protocols were carried out in accordance with the institutional guidelines of the University of Oviedo and were approved by the Animal Research Ethical Committee of the University of Oviedo prior to the study (approval code: PROAE 11/2014; date of approval: 9 December 2014). Experiments were done using female NOD.CB17/Prkdcscid/scid/Rj inbreed mice (Janvier Labs, St. Berthevin, France). For subcutaneous (s.c.) inoculations in the flanks of the mice, 5 <sup>×</sup> 105 cells mixed 1:1 with BD Matrigel Matrix High Concentration (Becton Dickinson-BD Biosciences, Erembodegem, Belgium) previously diluted 1:1 in culture medium were injected. Tumor volume was determined using a caliper as previously described [31]. One month after inoculation, mice were sacrificed by CO2 asphyxiation and tumors were extracted and processed for histological analysis. Limited dilution assays (LDA) and calculation of tumor-initiating frequencies (TIF) was performed as described in Supplemental Information. For the intra-bone (i.b.) inoculation, mice were anesthetized with isoflurane and the leg was bent 90◦ in order to drill the tip of the tibia with a 25G needle before cell inoculation (2 <sup>×</sup> 105 cells in 5 <sup>μ</sup>l of culture media per mouse) using a 27G needle [32]. In these experiments, tumor growth was evaluated in the preclinical image laboratory of the University of Oviedo using a computed tomography (CT) system (Argus CT, Sedecal, Madrid, Spain) at week 8 after cell inoculation and a μCT system (SkyScan 1174, Bruker, Antwerp, Belgium) following mice sacrifice at week 12 (see Supplementary Materials for analysis conditions).
