ATP1A1 is overexpressed in melanoma and is associated with a poor survival
Firstly, we analyzed a TCGA dataset using The Human Protein Atlas [28] and identified that ATP1A1 mRNA is significantly overexpressed in melanoma compared to 16 other cancer types (Figure S1). Subsequently, we assessed the protein expression of ATP1A1 using immunohistochemistry (IHC) in 84 skin and lymph node metastases (Table 1) and observed varying intensities of ATP1A1 staining on the cell membranes, which were ranked on a scale of 0 to 3 (Figure 1A). We then evaluated the percentage of tissue with such staining intensity to finally establish a score ranging from 0 to 100%. We searched for the optimal cut-off value by testing various percentiles from 20 to 80 by step of 5 and, dividing the population at 55% in low vs high ATP1A1 groups, we demonstrated that high ATP1A1 protein expression was significantly associated to shorter patient OS and, consequently, to a poor prognosis (Figure 1B).
ATP1A1 is predictive of resistance to BRAF inhibitor in patient samples
Furthermore, we investigated the expression of the ATP1A1 gene in 8 metastatic melanoma patients harbouring the V600EBRAF mutation, which is the most common mutation in melanoma. Importantly, these samples were obtained prior to treatment with the BRAF inhibitor vemurafenib. We observed a significantly higher expression of ATP1A1 mRNA in patients who did not respond to vemurafenib compared to those who showed a response (Figure 1C). Specifically, the expression of ATP1A1 was approximately 18-fold higher in tumor samples from non-responders compared to responders. Additionally, we examined the gene expression of other isoforms of the Na+/K+-ATPase α pump in these patients; however, we did not observe significant differences between responders and non-responders (Figure S2). These findings suggest that elevated ATP1A1 levels may serve as a specific predictive marker for resistance to BRAF inhibitor therapy.
High ATP1A1 level correlates with a differentiated phenotype both in patients and cell lines
To further characterize the role of ATP1A1 in melanoma, we conducted an analysis using the cBioPortal for Cancer Genomics (http://www.cbioportal.org/) based on data from The Cancer Genome Atlas (TCGA) of 448 cutaneous melanoma samples from the PanCancer Atlas. Our analysis revealed a positive correlation between high ATP1A1 mRNA expression and a panel of markers associated with differentiation and pigmentation, including MITF, TYR, TYRP1, TYRP2/DCT, RAB27A, SOX10, PAX3, and MLANA (Figure 1D) [29]. On the contrary, we observed a negative correlation between ATP1A1 expression and markers associated with a mesenchymal/invasive phenotype, such as AXL, EGFR, ZEB1, WNT5A, TGFβ1, SNAI1, MMP9, and TWIST2. These findings provide further evidence of the involvement of ATP1A1 in melanoma and its association with specific molecular markers related to cellular differentiation and pigmentation.
Additionally, examining the RNA-seq database and transcriptome profiling of 11 melanoma cell lines established by the Bordet team [30], we discovered that ATP1A1 is part of the Verfaillie proliferative gene signature, which is associated with the melanocytic state. This gene signature includes key transcription factors involved in melanocyte lineage, such as MITF and SOX10, as well as downstream markers like TYR, TYRP1, and MLANA, all of which play a role in cell differentiation and pigmentation. In contrast, ATP1A1 displays a strong negative correlation with the Verfaillie invasive gene signature (Figure 2A) (p<0.005). These findings collectively provide further evidence that ATP1A1 is predominantly expressed in differentiated and pigmented melanoma cell lines, consistent with the observations made in melanoma patients mentioned above.
High ATP1A1 expression is associated with acquired resistance to targeted therapy in cell lines
To further validate the involvement of ATP1A1 in resistance to BRAF-targeted therapy, we conducted experiments using three different models of melanoma cell lines (harbouring mutations in cKIT, BRAF or NRAS). Our findings consistently demonstrated a significant increase in ATP1A1 expression in cell lines that had acquired resistance to targeted therapies, respectively dasatinib, vemurafenib, or pimasertib, compared to their sensitive counterparts. This increase was observed at both the mRNA and protein levels (Figures 2B,C). Furthermore, among the isoforms of the Na+/K+-ATPase α pump, ATP1A1 was found to be the most highly expressed in both the sensitive and resistant MM074 and HBL cell lines, comparing the four isoforms (ATP1A1-4) (Figure 2D). It is worth noting that the MM161 and MM161-R cell lines also exhibited high levels of ATP1A3 expression. These results provide further evidence of the association between elevated ATP1A1 expression and resistance to targeted therapy, and highlight the predominant role of ATP1A1 among the isoforms in the tested cell lines.
ATP1A1 is colocalized with caveolin-1 in melanoma cell lines
To assess the subcellular localization of ATP1A1 in melanoma cells, we conducted immunofluorescence and confocal microscopy analyses. Our observations indicated that ATP1A1 predominantly localizes to the caveolae of cell membranes. Additionally, through dual immunofluorescence staining, we observed a significant colocalization of ATP1A1 with Cav-1, a marker protein for caveolae (Figures 3A,B). The physical interaction between ATP1A1 and Cav-1 was further confirmed using a proximity ligation assay, which demonstrated their close proximity and potential association (Figure 3C). These findings provide strong evidence for the specific localization of ATP1A1 in caveolae and its interaction with Cav-1 in melanoma cells.
Bufalin induced apoptosis of melanoma cells by acting on ATP1A1 in caveolae
Given the significant role of ATP1A1 in melanoma progression and resistance to targeted therapy, we investigated the impact of bufalin, a ligand of ATP1A1, on cell proliferation using crystal violet assays. Interestingly, the resistant melanoma cell lines exhibited higher sensitivity to the inhibitory effects of bufalin compared to their respective sensitive counterparts (Table 2). Furthermore, we identified a significant negative correlation between the IC50 values for bufalin and the mRNA levels of ATP1A1 across the six cell lines (rho=-0.943, p=0.005, Spearman correlation). To assess the effect of bufalin on colony formation, we conducted clonogenic assays and observed a dose-dependent reduction in colony numbers upon treatment with nanomolar concentrations of bufalin (Figure 4A). Moreover, we conducted flow cytometry analyses using Annexin V to evaluate the induction of apoptosis by bufalin. Remarkably, bufalin was found to induce apoptosis in both sensitive and resistant cell lines. Notably, the resistant cell lines, which expressed higher levels of ATP1A1, exhibited a more pronounced increase in apoptosis rate (Figure 4B). These results highlight the potential of bufalin to inhibit cell proliferation and induce apoptosis in melanoma cell lines, with a greater impact observed in the resistant lines that displayed elevated ATP1A1 expression.
To confirm the dependency of these effects on ATP1A1, we used siRNA to silence ATP1A1 expression in the BRAF-mutated MM074-R cell line. The knockdown efficiency was verified by qPCR, which demonstrated a significant reduction in ATP1A1 mRNA levels (Figure 4D). Subsequently, we evaluated the induction of apoptosis following bufalin treatment in cells transfected with ATP1A1 siRNA compared to cells transfected with scramble siRNA. The results revealed a notable decrease in apoptosis induction in ATP1A1 siRNA-transfected cells compared to the scramble siRNA-transfected cells (Figure 4C). This clearly indicates that the pro-apoptotic activity of bufalin is mediated through its targeting of ATP1A1.
Furthermore, to demonstrate that bufalin acts as an anti-cancer agent specifically when ATP1A1 is located in caveolae, we used methyl-β-cyclodextrin (MBCD) to disrupt caveolae formation. Remarkably, when cells were pre-treated with MBCD, there was a significant decrease in the anti-proliferative effect of bufalin (Figure 4E). These findings collectively demonstrate that bufalin-induced apoptosis is mediated by ATP1A1 in caveolae.
Bufalin affects the phosphorylation of Src and modulates downstream signalling pathways
Considering the localization of ATP1A1 in caveolae, we may expect an impact of the pump on the regulation of intracellular signalling pathways. Therefore, we investigated the effect of bufalin on signal transduction and specifically evaluated the phosphorylation profile of the Src protein, which is known to be downstream of ATP1A1. To assess these changes, we conducted Western blot analysis to examine the phosphorylation status of Src at Tyr416 (activation) and Tyr527 (inhibition) in cell lines treated with 10 nM bufalin for 24 hours. The results revealed that bufalin induced a decrease in Tyr416 phosphorylation in HBL, MM074, and MM161-R cells, indicating a reduction in Src activation. Conversely, bufalin treatment led to an increase in Tyr527 phosphorylation in MM074-R, MM161, and MM161-R cells, suggesting enhanced inhibition of Src activity (Figure 5A). These findings indicate that, with the exception of the HBL-R cell line, bufalin mediates its inhibitory signaling by reducing the activation phosphorylation and/or increasing the inhibition phosphorylation of the Src protein, resulting in Src inactivation.
To elucidate the signaling pathways affected by bufalin, we conducted a phosphokinase array to obtain a comprehensive overview of the phosphorylation changes induced by bufalin treatment in both sensitive and resistant cell lines (Figure 5B). The array encompassed the evaluation of protein phosphorylation and expression for a total of 39 proteins. We observed that bufalin treatment resulted in an overall increase in protein levels, with fold changes of 1.63 in HBL, 1.59 in HBL-R, 1.53 in MM074, 1.51 in MM074-R, 1.21 in MM161, and 1.29 in MM161-R. Notably, these increases showed a significant negative correlation with the IC50 values of the cell lines for bufalin (r=-0.943, p=0.005) (Table 2). This suggests that the most sensitive cell lines, namely HBL and HBL-R, exhibit the highest extent of phosphorylation changes in response to bufalin treatment. Importantly, the mRNA expression of ATP1A1 positively correlated with the overall levels of protein phosphorylation (r=0.886, p=0.019), and negatively correlated with the IC50 values for bufalin (r=-0.829, p=0.042).
Upon investigating the main changes induced by bufalin, we observed that in at least two cell lines, several proteins showed upregulation (defined as fold-change > 2) including akt1-2-3_S473, C-JUN, HSP27, P70-S6-KINASE_T421_S424, RSK1-2-3, and HSP60. These findings suggest that ATP1A1 is associated with a network involving AKT, JUN, RPS6K, and HSP (Figure 5C). Additionally, at least two cell lines exhibited downregulation (fold-change > 2) of FgR and STAT6. Notably, paired cell lines displayed discrepancies in the variations of certain proteins. For example, akt1-2-3_S473 and akt1-2-3_T309 were downregulated in HBL but upregulated in HBL-R, CREB_S133 showed an increase in HBL but a decrease in HBL-R, and PDGF_RB exhibited enhanced expression in MM074 while reduced expression in MM074-R. Moreover, ATP1A1 mRNA levels positively correlated with changes in GSK_3B (r=0.829, p=0.04), P53_S15 (r=0.829, p=0.04), and STAT6 (r=0.829, p=0.04), while the IC50 value for bufalin negatively correlated with variations in STAT6 (r=-0.943, p=0.005).
Altogether, although bufalin induced similar cellular effects such as cell apoptosis and inactivation of Src in all tested cell lines through ATP1A1 inhibition, it exerted its effects by influencing different signaling pathways. Furthermore, when comparing paired lines (sensitive vs resistant to targeted therapy), bufalin regulated different phosphoproteins. Additionally, no significant difference (according to the Mann-Whitney test) was observed when comparing sensitive and resistant cells. These findings further suggest that bufalin acts through cell-specific mechanisms, as these lines exhibit distinct phenotypes, alternative pathways, and metabolic switches [31].
Bufalin reduces tumour development in vivo
Finally, we conducted an in vivo study to assess the safety of bufalin and its impact on tumor development. In order to ensure a continuous distribution of bufalin and avoid potential issues related to its cardiotonic effects resulting from the serum peak after intraperitoneal (IP) administration, we used Alzet osmotic pumps to deliver a stable dose of 2 mg/kg/day of bufalin to the nude mice for a duration of 14 days. Specifically, focusing on the BRAF mutant model, we xenografted MM074 or MM074-R cell lines into mice to establish tumors that are either sensitive or resistant to targeted therapy (vemurafenib). Subsequently, the animals were exposed to bufalin for a period of 14 days (Figure S3, experimental design).
Regarding the toxicity of bufalin, we monitored the body weight of the mice throughout the treatment period and compared it between the control and treated groups. Importantly, we did not observe any significant changes in body weight between these two groups (Figure S4). Furthermore, at the time of sacrifice, we collected and weighed the organs from the mice in both the control and treated groups. Subsequently, we performed Haematoxylin-Eosin staining to examine the histological structures of these organs. Encouragingly, we did not observe any noticeable impact of bufalin on the organ/body weight ratio (Figure S4) nor on the morphology of the normal tissues (Figure 6A and Figure S5). Moreover, as bufalin is known to be a cardiotonic steroid that could potentially induce cardiac dysfunction as a side effect, we conducted troponin I staining in the hearts of control mice, bufalin-treated mice, and mice with confirmed cardiac dysfunction (used as positive controls). Importantly, we did not observe any troponin I staining in the hearts of both control mice and bufalin-treated mice, unlike the mice with documented heart dysfunction (Figure 6B).
About tumor progression in mice, we found a significant decrease in tumor volumes in the bufalin-treated group compared to the control group. Specifically, starting from day 6, we observed a significant reduction in tumor volumes in the bufalin-treated group for the sensitive tumors (Figure 6C). Similarly, for the resistant tumors, we observed a significant decrease in tumor volumes from day 9 in the bufalin-treated group compared to the control group (Figure 6D). Additionally, when evaluating tumor size and weight at the time of sacrifice (Day 14), we observed a significant decrease of the tumor size and weight in the bufalin-treated mice for both sensitive tumors (Figure 6E) and resistant tumors (Figure 6F) compared to not treated mice.