Differential gene expression analysis of cancer-related genes between BC cells
Published results of different trials show that the use of retinoids in BC patients has controversial results. Since RA exerts its effects via RARs and that Src-FAK are critical proteins involved in cancer progression, we initially explored the expression profiles of RARA, RARB, RARG, SRC, and PTK2 in different types of human BC cells (T-47D, MCF7, ZR75.1, BT-474, SK-BR3, MDA-MB-231, and MDA-MB-468) and a normal breast cell line (MCF-10A). For this purpose, we performed a transcriptome analysis using two datasets from the publicly available Gene Expression Omnibus database (GSE70884 and GSE68651). We found that RARA (RARα), SRC (Src), and PTK2 (FAK) are highly expressed in all breast cells. MDA-MB-231 and SK-BR3 have the lowest and highest level of RARA expression, respectively. RARB and RARG are expressed at similarly low levels in all breast cell lines (Fig. 1A).
We next compared the expression of sixteen genes between RA-resistant (MDA-MB-231 and MDA-MB-468), RA-sensitive (SK-BR3 and T-47D) BC cells, and a normal breast cell line (MCF-10A) using the previously obtained expression data. The selection of this set of genes was since they are biomarkers of different processes involved in BC metastasis such as MSN (moesin), CFL1 (cofilin) (actin cytoskeleton remodeling/depolymerization process); VCL (vinculin), TLN1 (talin), PXN (paxillin), DNM2 (dynamin) (FAs dynamics); CTTN (cortactin), WASL (family of the Wiskott Aldrich Syndrome Proteins, WASP), ACTR2 (Arp2 subunit), ACTR3 (Arp3 subunit) (actin nucleation); CDH1 (cadherin-1), VIM (vimentin), CDH2 (cadherin-2), CTNNB1 (β-catenin) (Epithelial-Mesenchymal Transition, EMT); and MMP2, MMP9 (matrix metalloproteinases 2 and 9) (invasion/extracellular matrix disassembly). All samples were sorted based on a hierarchical cluster algorithm with average linkage and Pearson’s correlation distance. According to Silhouette dendrograms, analysis samples were grouped into two major clusters. In one branch, it groups MDA-MB-231 and MDA-MB-468 (RA-resistant) cell lines; in the other, it groups SK-BR3, T-47D (RA-sensitive), and MCF-10A (normal) cell lines. Figure 1B shows genetic similarities between RA-resistant BC cell lines and, on the other hand, between RA-sensitive BC cell lines and the normal ones. Furthermore, we observed that the expression of VIM, MSN, TLN1, and CTNNB1 is markedly different between RA-sensitive and RA-resistant BC cells. In RA-resistant cells, these genes are overexpressed; meanwhile, in RA-sensitive are underexpressed. We noted that the expression of these genes is similar between RA-sensitive cells and the normal breast cell line MCF-10A (Fig. 1B, genes marked in blue). MMP9 expression is low in RA-resistant compared with RA-sensitive and normal lines (Fig. 1B). Additionally, we found that the expression of CTTN and CDH2 is overexpressed in RA-resistant and normal breast cells compared to RA-sensitive (Fig. 1B, genes marked in orange). Altogether, this suggests that VIM, MSN, TLN1, CTNNB1, CTTN, MMP9, and CDH2 are promising candidates to be studied as markers of RA response.
To further elucidate whether genes are associated with RA-resistance, we performed a differential gene expression analysis in control and RA-treated MDA-MB-468 and MDA-MB-231 BC cells. In Figure 1C-D, both cells show the same behavior. RA administration (100 nM, 18 h) modulated the expression of all the genes analyzed, except ACTR2 and CFL1. Genes like WASL, CTTN, SRC, and DNM2 were upregulated (Fig. 1C-D, genes marked in red), while VIM, MSN, TLN1, PXN, VCL, ACTR3, and PTK2 were downregulated (Fig. 1C-D, genes marked in green). These results suggest that RA-resistance would reflect deregulation of most RA-target genes, mainly those encoding components from the signaling of the Src-FAK pathway.
Next, based on SK-BR3 and T-47D having a high genomic similarity concerning metastatic BC patients [34], we examined the differential gene expression between SK-BR3/T-47D and normal mammary cell line MCF-10A. We used a gene expression dataset from the publicly available Gene Expression Omnibus database (GSE103426). The volcano diagrams show that both BC cells exhibit a decreased expression of VIM, CDH2, and MSN compared to MCF-10A (Fig. 1E-F, genes marked in green). We also found that SK-BR3 has an increased expression of RARA, WASL, SRC, and MMP9 compared to MCF-10A (Fig. 1E, genes marked in red), while T-47D has an increased expression of CDH1 and PXN (Fig. 1F, genes marked in red).
RA reduces cell adhesion and migration in tumoral cells
To confirm the role of RA in inhibiting tumor progression, we first verified the expression of RARs, Src, and FAK in T-47D, SK-BR3, and LM3 by western blot analysis. We found that RARα, RARβ, RARγ, Src, and FAK are present in all BC cell lines analyzed (Fig. 2A).
Then, we evaluated the RA effect on cell adhesion and migration in human T-47D and murine LM3 BC cell lines. In both models, treatment with RA (1 - 10 µM) for 72 h markedly decreased cell adhesion to an extracellular matrix-like surface by 58-79%, respectively, compared to the control group (Fig. 2B-E). Likewise, the administration of RA (1 - 10 µM) during 72 h reduced cell migration by 52-88% in T-47D and by 25-35% in LM3 cells (Fig. 2H-K). Furthermore, we noted that RA inhibitory effects are extrapolated to other cancer cells. Treatment with RA (1 - 10 µM) for 72 h in human cervical carcinoma cell line HeLa significantly inhibited cell adhesion (Fig. 2F-G) and migration (Fig. 2L-M) in a dose-dependent manner.
Treatment with RA plus FAKi reduces RARα and FAK expression and decreases LM3 cells viability
Due to the fundamental role of RA and FAK in tumor progression, we next evaluated whether the combined treatment of RA with the specific inhibitor of FAK (FAKi) improves the inhibition of tumorigenesis. First, we tested the sensitivity of LM3 cells to drugs. We performed an MTT assay using a dose range of RA (0.01 – 100 µM) and FAKi (0.5 - 2 µM) and their combinations for 72 h. We observed that treatment with RA decreased cell viability in a dose-dependent manner. The IC50 was reached with a dose of 100 µM RA (Fig. 3A). The maximal dose tested of FAKi showed a significant decrease in cell viability of 38% compared with untreated cells. A synergistic effect was observed in the inhibition of cell viability when combined treatments were administered at the highest concentration (Fig. 3A). These results support the concept that LM3 BC cells are sensitive to RA and FAKi treatment.
Next, LM3 cells were treated for 72 h with RA (1 µM), FAKi (1 - 2 µM), and their combinations to analyze the expression of RARs and FAK (Fig. 3B). We found that all treatments induced downregulation of RARα and FAK compared to the control. The RA+FAKi combinations resulted in higher downregulation than RA alone (Fig. 3B-C and F). On the other hand, we observed that the administration of RA induced upregulation of RARβ, while the treatment with FAKi (2 µM) decreased its expression compared to the control group. Moreover, RA+FAKi reduced RARβ expression compared to RA but did not affect its expression compared to the control (Fig. 3B and D). Regarding RARγ, we observed that RA and FAKi reduced its expression as a single agent compared to the control; however, their combinations did not affect it (Fig. 3B and E).
We also performed an immunofluorescence assay on LM3 cells treated with RA (1 µM), FAKi (1 µM), and their combination for 72 h to reveal the expression and subcellular localization of FAK. We visualized that FAK is homogeneously and diffusely localized throughout the cytoplasm in control cells, whereas treatment with RA and RA+FAKi triggers FAK nuclear relocalization (Fig. 3G). Furthermore, we observed in all treatments the longitudinal arrangement of actin fibers, typical characteristics of a static cell phenotype, suggesting that drug administration affects the reorganization of the actin cytoskeleton (Fig. 3G), preventing the cell from being able to drive its motor machinery impairing BC cell motility. In parallel, we also confirmed that the combined treatment significantly reduced FAK expression (mean intensity, pixel/area) (Fig. 3H).
RA and FAKi combination improve LM3 adhesion and migration inhibition
We tested the ability of LM3 cells to adhere to an ECM substrate after RA (1 µM), FAKi (1 µM), and their combination during 72 h. We showed that treatment with RA, FAKi, and RA + FAKi reduced cell adhesion compared to the control. The RA plus FAKi produced a more significant inhibition than RA alone (Fig. 4A-B). In parallel, we exposed cells to RA, FAKi, and RA+FAKi and we monitored cell motility for 72 h. We found a significant inhibition of cell migration in all conditions tested compared to the control, reaching the major inhibition of migration in the combined treatment (Fig. 4C-D).
RA plus FAKi reduces tumor growth and metastasis, increasing mice survival
Finally, we performed a murine in vivo model to determine correlation with the in vitro results. Tumor growth was evaluated by the orthotopic tumor growth assay consisting of inoculating murine LM3 cells, treated or not with FAKi (1 µM, 72 h), in the mammary gland BALB/c mice bearing or not RA (10 mg)-subcutaneous pellet. We observed that RA and FAKi administered separately reduced the tumor growth, but the combined treatment induced a more potent inhibition (Fig. 5A-B). In concordance, RA and FAKi treatment increased mice survival, but only the combination RA+FAKi was statistically significant (Fig. 5C). In addition, we observed that in controls and in the FAKi group the tumors were histopathologically aggressive, even ulcerating the dermis. The treatment with RA and the combination RA+FAKi prevented the development of these highly proliferative tumors. Representative images of ulcers found in control animals (Fig. 5D-E and H) and tumors of combined treatments are shown (Fig. 5F-G). The total body weight of mice was not different in the experimental groups compared to control, demonstrating a lack of RA-dependent toxicity (data not shown).
On the other hand, since Ki-67 is widely used in routine pathology as an established biomarker of cell division, to assess the proliferation rate of human BC tumors, we measured Ki-67 expression by immunohistochemistry in mice tumors. We found that the percentage of positive cells for Ki-67 was markedly lower in the treatments with FAKi and RA+FAKi compared to the control group (Fig. 5I-J).
Finally, we performed an experimental metastatic assay to analyze the effect of the RA+FAKi combination on the spread of tumor cells. LM3 cells, pretreated or not with FAKi (1 µM, 72 h), were injected into the tail vein of mice bearing or not RA (10 mg)-silastic pellet. We found that RA significantly reduced metastatic lung dissemination. FAKi and the RA+FAKi combination presented a lower but non-significant number of lung nodules than the control group (Fig. 5K-L).