Digitoxin Affects Metabolism, ROS Production and Proliferation in Pancreatic Cancer Cells Differently Depending on the Cell Phenotype

Digitoxin has repeatedly shown to have negative effects on cancer cell viability; however, the actual mechanism is still unknown. In this study, we investigated the effects of digitoxin (1–100 nM) in four pancreatic cancer cell lines, BxPC-3, CFPAC-1, Panc-1, and AsPC-1. The cell lines differ in their KRAS/BRAF mutational status and primary tumor or metastasis origin. We could detect differences in the basal rates of cell proliferation, glycolysis, and ROS production, giving the cell lines different phenotypes. Digitoxin treatment induced apoptosis in all four cell lines, but to different degrees. Cells derived from primary tumors (Panc-1 and BxPC-3) were highly proliferating with a high proportion of cells in the S/G2 phase, and were more sensitive to digitoxin treatment than the cell lines derived from metastases (CFPAC-1 and AsPC-1), with a high proportion of cells in G0/G1. In addition, the effects of digitoxin on the rate of glycolysis, ROS production, and proliferation were dependent on the basal metabolism and origin of the cells. The KRAS downstream signaling pathways were not altered by digitoxin treatment, thus the effects exerted by digitoxin were probably disconnected from these signaling pathways. We conclude that digitoxin is a promising treatment in highly proliferating pancreatic tumors.


Introduction
Pancreatic ductal adenocarcinoma (PDAC) is a tumor in the exocrine part of the pancreas [1]. It has one of the poorest prognoses of all cancer forms, with a five-year survival rate of less than 5% [2]. The currently used therapies have considerable adverse side effects on unspecific targets, including healthy normal cells [3]. Hence, new treatment regimens specifically affecting cancer cells are urgently needed. There are differences between normal cells and cancer cells that are distinct and common for all cancers, such as mutations in oncogenes and tumor suppressor genes, metabolic alterations, aberrant production of reactive oxygen species (ROS), and enhanced proliferation [4,5]. Normal cells have a strictly regulated proliferation controlled by mitogenic signaling [6]. Cancer cells with mutations in oncogenes and tumor suppressor genes proliferate at a high rate, without any mitogenic signaling [7]. Cancer cells also have an increased rate of metabolism with subsequent high ROS production, which affects the proliferation rate [8].
Digitoxin is a well-known cardiac glycoside [9][10][11] with the potential to act as a potent anti-cancer drug in the therapeutic concentration range of 25-40 nM [10,12,13]. Epidemiological data suggest that patients who received treatment with the cardiac glycosides digoxin or digitoxin have a higher survival rate and lower recurrence in different cancer malignancies [14,15]. Cancer cells seem to be more sensitive to digitoxin compared with normal cells, but the reason for this is still unknown [14]. Cardiac glycosides inhibit the Na + /K + -ATPase by binding to the subunit α, blocking its function. Increased intracellular Here, we used four well-characterized pancreatic cancer cell lines with a different mutational status and origin (primary tumor or metastasis) to cover as many variations as possible in PDAC, so as to study the effects of digitoxin. There were major differences in the basal phenotypes, metabolism, ROS production, and proliferation pattern in the four cell lines. The basal proliferative phenotype was important for the response to digitoxin treatment. Highly proliferating cancer cells were more vulnerable to the toxic effects of digitoxin. The effects seemed to be exerted via the increase in calcium and the resulting effects on the ROS concentration and cell proliferation. In our study, the signaling pathways PI3K/Akt and Raf/MEK/ERK1/2 were not affected by digitoxin treatment.

Digitoxin Treatment Decreased Pancreatic Cancer Cell Viability and Induced Apoptosis
Treatment with digitoxin at concentrations between 10-100 nM during 48 h incubation had negative effects on cell viability in all four cell lines ( Figure 1A−D). The sub-therapeutic concentration of 10 nM digitoxin resulted in a decrease in the number of viable cells compared with control cells, ranging between 12.6% (±2.1, p < 0.001) in Panc-1 ( Figure 1C) to 11.9% (±2.1, p < 0.001) in BxPC-3 ( Figure 1A) compared with the respective controls. In the therapeutic range (25-40 nM), the number of viable cells was more decreased, which resulted in between 42.8% (±2.0, p < 0.001) in Panc-1, 40 nM ( Figure 1C) to 22.8% (±1.8, p < 0.001) in CFPAC-1 for 25-40 nM ( Figure 1B). The digitoxin induced a reduction in the number of viable cells, with 9.7% in the AsPC-1 treated with 25 nM digitoxin compared with the control (±1.2, p = 0.01) and 22.0% in the samples treated with the highest concentration of digitoxin (100 nM) (±1.9, p < 0.001). In parallel with these results, there was an increase in apoptotic cells in the samples treated with digitoxin ( Figure Table S1).

Oncogenic Signaling in the PDAC Cell Lines
In order to analyze the basal PI3K/Akt and/or Raf/MEK/ERK activity in the cell lines and the possible effects of digitoxin treatment, we evaluated the protein levels of pAkt, Akt, pErk1/2, and Erk1/2 after 48 h of treatment and controls ( Figure 5A-H). The Western blot analysis showed no significant change in the protein expression after digitoxin treatment, except for the total Akt levels in CFPAC-1, and there was a significant increase in the Akt-level with increasing the concentrations of digitoxin (F1,4, p = 0.002) ( Figure 5B). Only Panc-1 had detectable levels of phosphorylated Akt (S473) (pAkt), but the protein level was very low and was not affected by digitoxin treatment ( Figure 5C). ERK1/2 was expressed in all four cell lines, and the active form, pERK1/2, was detectable in CFPAC-1, Panc-1, and AsPC-1 ( Figure 5F-H); however, the pERK1/2 and ERK1/2 levels were not significantly affected by digitoxin treatment in any of the cell lines. The total level of both Akt and ERK1/2 was higher in CFPAC-1 and Panc-1 compared with the other cell lines (Supplementary Figure S2).

Discussion
We have focused on the potential use of an already established cardiotonic drug, digitoxin, for the purpose of treating pancreatic cancer. Digitoxin has fatal effects on cell viability and it induced apoptosis to different degrees in the four pancreatic cancer cell lines examined in this study. These results are in accordance with previous studies using cardiac glycosides in human glioma and pancreatic cancer cell lines [49,50]. In an earlier study, we found intracellular calcium was significantly increased in the cell lines BxPC-3 (120%), CFPAC-1 (117%), and Panc-1 (160%) treated with digitoxin at therapeutic concentrations of 25-40 nM. In AsPC-1 cells, only the supra-therapeutic concentration, 100 nM, induced a significant increase in intracellular calcium with 12% (unpublished data) [51]. An increase in the intracellular Ca 2+ concentrations affects cell metabolism, proliferation, and apoptosis in cancer cells [52,53]. The four cell lines used in this study had major differences in their basal phenotype, which seemed to affect and dictate the response to digitoxin treatment. BxPC-3 and Panc-1, both derived from primary human tumors, were most affected by digitoxin treatment regarding their negative effects on cell viability. CFPAC-1, derived from a liver metastasis, and AsPC-1, from an ascites metastasis (mouse xenograft), were less affected by digitoxin. In contrast with the general negative effects of digitoxin on cancer cell viability, we detected differences in metabolism, ROS production, and proliferation in the PDAC cell lines [54,55].
The four investigated cell lines differed in their mutations in KRAS or BRAF, potentially affecting the downstream signaling pathways of PI3K/Akt and/or Raf/MEK/ERK [56][57][58][59]. CFPAC-1 with the KRAS p.G12V and BxPC-3 with BRAF mutation had a relatively high rate of glycolysis under basal conditions and also a high transcriptional expression of GLUT1, HK2, and PKM2. These genes are often found to be highly expressed in primary pancreatic tumors with the KRAS p.G12V mutation [46]. Fritsche-Guenther et al. concluded that colorectal carcinoma cells with the KRAS p.G12V mutation are highly glycolytic [60]. According to Yun et al., colorectal cancer cells with KRAS and BRAF mutations have a constitutively activated GLUT1 and enhanced glycolysis, in contrast with cells with mutations in PIK3CA (the catalytic part of PI3K) [44]. Panc-1 and AsPC-1 with the KRAS p.G12D mutation, had a lower basal rate of glycolysis and a low transcriptional expression of the genes important for glycolysis and glucose transport.
With digitoxin treatment, the BxPC-3 and CFPAC-1 cells maintained the rate of glycolysis, while the glycolysis rate decreased in Panc-1 and AsPC-1. However, digitoxin treatment did not affect the phosphorylation of Akt (PI3K/Akt pathway) or ERK1/2 (Raf/MEK/ERK1/2 pathway) in the cell lines in this study. Therefore, we conclude that the changed glycolysis rate is controlled via other pathways.
The high glycolytic activity in BxPC-3 and CFPAC-1 could be the reason for the relatively low increase in intracellular Ca 2+ in these two cell lines. Maintaining a low concentration of intracellular Ca 2+ is fundamental for the normal function of the cell and is mainly managed by PMCA [61]. As shown in a study by James et al., 2020, PMCA function depends on ATP produced by glycolysis in PDAC [52].
In contrast, Panc-1 with its low glycolysis rate was not able to provide PMCA with enough ATP, hence the large increase of intracellular Ca 2+ . Mitochondria are also important in the regulation of calcium homeostasis [23,62,63]. When the intracellular concentrations of Ca 2+ increase, some of the excess Ca 2+ is transported into the mitochondria, causing upregulation of the citric acid cycle and oxidative phosphorylation [62]. If the intracellular Ca 2+ increases too much, it will induce apoptosis [64].
Digitoxin treatment caused an increase in ROS production in all four cell lines, but to the highest degree in BxPC-3 and Panc-1. As oxidative phosphorylation is the major producer of ROS, we hypothesized that the increase in ROS was due to the increased activity of oxidative phosphorylation caused by Ca 2+ . ROS is important for controlling the redox state of the cell and, further, to control the cell cycle [26,28,31]. We found that BxPC-3 and Panc-1 have a high proportion of cells in S/G2 phase. Cells in this phase are more dependent on oxidative phosphorylation than cells in the other phases in the cell cycle [29]. The large proportion of cells in S/G2 in Panc-1 make this cell line highly dependent on oxidative phosphorylation for energy and molecules for nucleotide and lipid synthesis [65]. Panc-1 also had an increased proportion of cells in S/G2 with digitoxin treatment, which was reflected in the increase in oxidative phosphorylation (ROS) and decrease in the glycolysis rate. An increase in the proportion of cells in the S/G2-phase might imply DNA damage. In a study on HeLa cells by Gan et al., 2020, they found digitoxin to cause cell arrest in G2 due to DNA double strand breaks [66].
Both BxPC-3 and CFPAC-1 had an increase in cells in the G0/G1 phase when treated with digitoxin. These cell lines were similar in their basal rate and response in glycolysis rate to digitoxin treatment, and during treatment, they increased the proportion of cells in G0/G1. The increased proportions of cells in G0/G1 in BxPC-3 and CFPAC-1 with digitoxin treatment could be due to an arrest in G1. In earlier studies, cardiac glycosides have been shown to introduce arrest in G1 both in non-small cell lung cancer and in large cell lung cancer [67,68]. As the cells in the G1 phase are dependent on energy and products from glycolysis, this is a feasible explanation for the maintained rate of glycolysis in BxPC-3 and CFPAC-1 [69].
AsPC-1 with KRAS p.G12D was outstanding because of the exceptionally low effects of digitoxin treatment compared with the other cell lines. AsPC-1 had a low basal rate for both glycolysis and oxidative phosphorylation, which could reflect a high proportion of cells in G0/G1. We believe that most of the cells in AsPC-1 were in the quiescent G0 phase, with only basal oxidative metabolism. They were slow growing and had a longer doubling time compared with the other cell lines, which supports the idea of a quiescent mode. Invading cancer cells are quite insensitive to digitoxin and have most of their cells in G0 [70]. The high chemoresistance found in invading cancer cells was probably due to the fact that they are mainly in the G0/G1 phase [27].
We found large differences in response to digitoxin in the cell lines in this study. BxPC-3 and CFPAC-1 seemed to be halted in G1, Panc-1 in G2 in the cell cycle, while AsPC-1 was unaffected. Wang et al., 2021, also found large differences in response to digoxin treatment between two cell lines derived from non-small cell lung cancer (NSCLC)-one of the cell lines was arrested in G1 and the other in G2 [68].
We therefore conclude that the beneficial effects of digitoxin on pancreatic cancer cells are the most pronounced in primary, highly proliferating and non-migrating tumors. In addition, when using cell lines as model systems in cancer research, it is important to be aware of the phenotypic/genotypic differences between the cell lines.

Seeding and Treatment with Digitoxin
All of the analyses were performed in 96-well plates and after 48 h of treatment with digitoxin (Sigma-Aldrich, St. Louis, MO, USA), 5000 cells were seeded in 100 µL complete growth medium and incubated at 37 • C with 5% CO 2 to a sub-confluent monolayer. After 20 h of incubation, the media were removed and 100 µL new media containing digitoxin were added to the cells. The control cells only received new media. The concentrations of digitoxin used were 0 nM (controls), 1 nM, 10 nM, 25 nM, 40 nM, and 100 nM (human therapeutic range, 25-40 nM). Cells were further incubated at 37 • C with 5% CO 2 for 48 h.

RNA-Extraction, cDNA Synthesis, and Real-Time PCR
For each RNA extraction, 1 × 10 5 cells were seeded in 6-well plates in 2.4 mL complete medium, and they were incubated for 20 h at 37 • C with 5% CO 2 to a sub-confluent monolayer. After 20 h, the media were removed and new media were added together with digitoxin in the same concentrations as previously mentioned. The control cells only received new media.

Cell Viability Assay
Cell viability analysis was performed with the CellTiter 96 ® AQ ueous One Solution Cell Proliferation Assay (MTS) (Promega, Madison, WI, USA). This is a colorimetric method for determination of the number of viable cells. After 48 h of incubation with digitoxin, 20 µL of MTS tetrazolium was added to each well and they were then incubated for 1 h. The quantity of the formazan product was measured using a spectrophotometer at 490 nm (FLUOstar Omega, BMG Labtech, Ortenberg, Germany). The value was directly proportional to the number of living cells in the cell culture. All of the assays were performed in octuplicates and were repeated five times (Table S1: MTS).

Caspase 3/7 Assay
Cells were seeded in black-walled 96-well cell culture plates and were treated with digitoxin, as previously described, and incubated for 48 h. The endpoint assay in the CellEvent™ Caspase-3/7 Green Detection Reagent kit (C10423, Invitrogen/ThermoFisher Scientific, Waltham, MA, USA) was performed. Reagents in this assay consisted of a DEVD peptide conjugated to a fluorescent dye binding to nucleic acids. The activation of Caspase 3 and/or Caspase 7 cleaved the conjugated DEVD peptide and the nucleic acid binding dye became fluorescent when binding to the DNA. The fluorescence intensity was measured with a microplate reader, FluoStar, at Ex/Em 485/520 (FLUOstar Omega, BMG Labtech, Ortenberg, Germany). The assay was performed in triplicate and repeated three times.

Extracellular Lactate-Aerobic Glycolysis
Cells were seeded in 96-well plates and after 48 h of treatment with digitoxin, 5 µL of media was diluted 1:20 in PBS and kept in -20 • C until the analysis. The samples were further diluted to a final dilution of 1:800 and were analyzed in by Lactate-Glo™ Assay (Promega, Madison, WI, USA). Luminescence was measured in a microplate reader, FluoStar. The assay was performed in duplicate and was repeated twice.

Intracellular ROS
Cells were seeded and treated with digitoxin in triplicate for 48 h, as described earlier.
The oxidative stress was measured using the CellROX green reagent (C10444, Invitrogen/ThermoFisher Scientific, Waltham, MA, USA). The dye has a bright green fluorescence when oxidated by ROS and after subsequent binding to DNA. The fluorescence intensity was measured with a microplate reader, Fluostar, at Ex/Em = 485/520 (FLUOstar Omega, BMG Labtech, Ortenberg, Germany). The assay was performed for two biological replicates and two technical replicates.

Cell Cycle Analysis
The Cell-Clock assay (Biocolor, Carrickfergus, UK) was used to assess the basal proliferation status and the effects of digitoxin on the cell cycle in the four cell lines. This is a system to measure the four phases, G0-G1, S, G2, and M, of the cell cycle in live cells. Cells were seeded in 96-well plates, treated with digitoxin for 48 h, and incubated with the provided redox dye for 1 h at 37 • C; then, they were manually evaluated by two researchers blinded to the actual treatment. The ratio of each phase was calculated (Table S2: Cell cycle).

Statistical Analysis
Statistical analysis was performed using IBM SPSS Statistics 27.0 (Armonk, NY, USA: IBM Corp), Student's t-test (paired, two-tailed) for cell viability, caspase 3/7, and Cell-Clock results to confirm significance among treatments, compared with the control. The effect of digitoxin treatment on the transcriptional expression, protein expression, and lactate production was assessed by linear regression. Significance values were * p ≤ 0.05, ** p ≤ 0.01, and *** p ≤ 0.001.

Conclusions
The high malignancy and poor prognosis of PDAC, together with increasing patient numbers, call for the development of new treatment strategies. Increased knowledge about the effect that digitoxin exerts on PDAC cells is needed to validate its use for the treatment of pancreatic cancer. In order to obtain a maximal effect on cancer cell viability, a combination of treatments affecting metabolism and proliferation in cancer is preferable.
We have shown that digitoxin can act as an anti-cancer agent in several pancreatic cancer cell lines, but it affects metabolism, ROS production, and proliferation differently depending on the cell type. Cell metabolism and proliferation are closely connected because of the changing demands in the different phases in the cell cycle, and the proportion of cells in the S-phase seems to determine the success of digitoxin treatment.
Our results show that BxPC-3 and Panc-1, derived from primary tumors, have a high proportion of cells in the building/synthesizing phases in the cell cycle, and are also the most vulnerable to digitoxin treatment. Knowing the metabolic and proliferative pattern in human tumors will be critical for choosing the correct treatment.