br The AKT pathway is

The AKT pathway is also known to crosstalk with the Notch signaling system [4]. A phosphorylation-specific antibody array for the activation of AKT pathway (Cell Signaling Technology) showed less phosphorylation of various AKT-related molecules in the COMP-expressing Iberiotoxin compared to control cells (data not shown). Reduced phosphorylation of AKT at serine 473 and threonine 308 was observed in COMP-expressing as compared to control cells (Fig. 5H, I). This effect was reversed by inhibiting Notch signaling using DAPT (Fig. 5H, I). The reduction in AKT phosphorylation was independent of PTEN, an inhibitor of PI3K, since the total PTEN protein levels were not altered, even after DAPT treatment (Fig. 5H). Also, the activation of ERK1/2 remained unchanged in the presence of COMP. Taken together, COMP yields activation of β-catenin, which is also regulated by the activation of Notch3, and deactivates AKT independently of PTEN.
COMP knock out mice fail to develop cancer stem cells
To characterize the effect of COMP on the population of cancer stem cells in vivo, we intercrossed the 
spontaneous breast cancer forming MMTV-PyMT mice with the dominant negative COMPKO mice [33]. Tumor volumes measured in COMP+/+ mice were significantly larger compared with COMP−/− mice, with the COMP−/+ showing similar tendency as the full knock-out (Fig. 6A). No difference was observed in the number of lung metastases (Fig. S3B). The cancer stem cell population in excised tumors was detected using flow cytometric analysis of 5 cancer stem cell markers (CD24, Thy1 (CD90), CD133, CD61, CD29). Cells positive for the markers CD24 and Thy1 (CD90) were previously defined as cancer stem cells in the MMTV-PyMT mouse model [36]. COMP+/+ tumors contained a significantly larger population of cancer stem cells compared with the COMP−/− tumors, p b 0.0001 as well as with the heterozygote tumors, p = 0.0007 (Fig. 6B, C). Furthermore, a portion of the tumor cells was lysed, and total proteins were analyzed with western blot immunodetection of Notch3. Activa-tion of Notch3 was evident in 6 out of 7 total analyzed tumors from COMP+/+ mice, of 6 total COMP−/− mice, and of 4 total COMP−/+ mice (Fig. 6D). Moreover, the NICD3 band intensity was correlated with the propor-tion of cancer stem cells presented in the tumors, using the Spearman's rank correlation coefficient ps = 0.0043 and r2 = 0.7 (Fig. 6E). r> Discussion
During the last decades, our understanding of the factors governing tumor organization both in vitro and in vivo has improved. The view has shifted from the initial hypothesis that all tumor cells have the same characteristics with regard to self-renewal and metastatic dissemination [37] to a theory that tumors are well-organized groups of cells with a certain hierarchy, where only a small percentage of cells has the ability to give rise to different types of tumor cells, to metastasize, and to self-renew. An improved understanding of cancer stem cell biology might lead to the development of drugs specifically targeting this population of cells [38]. In this study, we propose a new molecular mechanism for Notch3 and Jagged1 interaction mediated by the extracellular matrix protein COMP, regulating the proportion of cancer stem cells.
In a previous study, we showed that xenografts of MDA-MB-231 cells expressing COMP formed larger tumors than control cells [25]. To dissect the underlying molecular mechanism in the framework of a genetically controlled in vivo system, we resorted to the well-established MMTV-PyMT trans-genic model for breast cancer and crossed this mouse into a COMP-deficient genetic background [33]. We observed that the tumor volume was larger in the COMP+/+ background compared to both the COMP−/− and COMP+/− backgrounds and thus, the reduction of COMP (in heterozygous animals) had a
116 COMP leads to activation of Notch3
Fig. 5. Notch3 activation, by the plasma membrane rebound COMP, orchestrates the cross-talking of main cancer stem cell molecular pathways. (A) Cell culture medium of MDA-MB-231 cells was supplemented with recombinant COMP or BSA as a control, (Neg., no protein added) and incubated overnight. Cells were lysed and Notch3 activation was assessed using western blot analysis with β-tubulin serving as a loading control. (B) Supplementation of cell culture medium with 80 μg/ml of recombinant COMP induced activation of Notch3 protein levels. (C) Jagged1 levels evaluated by western blot analysis in cell lysates from MDA-MB-231 cells incubated with recombinant COMP or BSA overnight (D) Addition 40 μg/ml and above of recombinant COMP in cell culture medium lead to the reduction of total Jagged1 protein levels. (E) Luciferase reporter assay measuring β-catenin activity. (F) Western blot analysis in total cell lysates. Both methods (E & F) were performed using MDA-MB-231 cells expressing COMP, the corresponding control cells and COMP expressing cells incubated with DAPT for 24 h. (G) Cytoplasmic and nuclear fraction were isolated from MDA-MB-231 cells and western blot analysis was used to detect total and active β-Catenin. Lamin and GAPDH served as loading controls for the nuclear and cytoplasmic fragments, respectively. (H) Representative western blots of total MDA-MB-231 cells lysates expressing COMP protein and correspondingly Notch3 activation inhibited utilizing DAPT for 72 h. (I) Densitometry of AKT phosphorylation western blot analysis. All experiments were repeated at least 3 times. One-way ANOVA, Bonferroni's multiple comparisons test was used when compared 3 or more samples, and Two-way ANOVA Bonferroni's multiple comparisons test was used when comparing 3 or more groups with 2 variables (*b0.05, **b0.01, ***b0.001, ****b0.0001). NICD: Notch Intracellular Domain, NEXT: Notch Extracellular Truncated, FL: Full-length protein.