Briefly, HepG2 cells which were cultured in 10% FBS-supplemented RPMI 1640 medium, seeded (5 × 10³) in 96-well culture plate. After 24 h in CO₂ incubator (37 °C), cells were incubated for 72 h with serial concentrations of DF(II) NPs, DF(II), DE, F(II)NPs, and FeO. According to Mosmann (1983), MTT assay was used to assess the percentage of growth inhibition in the treated cells relative to the untreated cells and then the dose at which 50% growth inhibition (IC₅₀) was estimated using GraphPad Prism 9. Additionally, morphological alterations in chelating complexes (the most active compounds)-treated HepG2 cells were recorded using phase contrast inverted microscope (Olympus, Japan).

Briefly, HepG2 cells were treated with the above-mentioned compounds (at the lowest IC₅₀ values “3.4 μg/ml”). After 72 h, the elevation in cellular content of lipid peroxidation was determined in the supernatant, relative to the untreated cells, using thiobarbituric acid reactive substance method (Girotti et al., 1991).

After 72 h treatment of HepG2 with 3.4 μg/ml of DF complexes (DF(II) NPs and DF(II)). Cell death can be observed after staining the untreated and treated cells with acridine orange/ethidium bromide (AO/EB) using fluorescence microscope (Olympus, Japan). For accurate quantification, cells were stained with fluorescein isothiocyanate (FITC)-labelled annexin V/propidium iodide (PI) and analyzed at FITC signal detector (FL1) and phycoerythrin emission signal detector (FL2) using flow cytometry (Partec, Netherlands) with FloMax software.

The anti-migration activity of DF complexes was evaluated by 24 h incubation with scratched monolayer HepG2. The wound closure area was recorded, by a digital camera-phase contrast inverted microscope, at 0 h and 24 h of treatment and then analyzed using ImageJ NIH software.

Typical procedures were used to prepare the tissue slides for the histological investigation of liver, lung, and spleen by H&E staining and immunohistochemical assessments (ki-67 immunostaining tumor tissues “liver and lung” as well as N-cadherin immunostaining liver). CellSens and ImageJ were used to analyze the immunostaining images.

The level of AFP (a basic liver tumor marker) in the blood was quantified using the AFP-chemiluminescent immunoassay kit.

In a separate animal study, LC-bearing mice (b.w ∼ 20 g, n = 15) were divided into three groups (untreated LC, LC-DF(II) NPs, and LC-DF). The two treated LC groups were received intraperitoneally 0.04 mg of DF complexes per day for 1 week. After that, mice were sacrificed and tissues (hepatic nodules, lung, spleen, brain, heart, and kidney) were harvested. Using graphite atomic absorption spectrometry (Analytik Jena AG, Germany), the tissue level of iron which is equivalent to the accumulative tissue uptake of DF complexes, was quantified after normalization by subtracting from the corresponding iron tissue level of the untreated LC group.

The following redox parameters were assessed in liver and lung tissue homogenates of the treated LC groups, relative to the untreated groups, after centrifugation at 3,000 g for 15 min at 4 °C. The fold increment in ROS and lipid peroxidation was determined using DCFD (Hempel et al., 1999) and TBA reactive substance (Ohkawa et al., 1979) methods, respectively. Furthermore, the relative fold decrement in antioxidant parameters (Nrf2 transcription activation and GSH content) were measured in these tumor tissue homogenates according to the instructions of Nrf2 transcription factor assay kit and Ellman’s assay (Ellman, 1959), respectively.

Briefly, following the tissue homogenization in 0.1 M Tris-HCl pH 7.4 (0.25 M sucrose, 5 mM β-mercaptoethanol, and protease inhibitor cocktail) and centrifugation, GPX4 activity was measured using coupled enzyme assay (Roveri et al., 1994). To begin reaction, tissue samples were added to the mixture of 0.14 U glutathione reductase, 0.2 mM NADPH, 3 mM GSH, 0.1 mM cumene hydroperoxide (substrate) and then oxidation of NADPH was recorded, per min, at 340 nm (Kernstock & Girotti, 2008). Regarding ALDH2 activity, in brief, the tissue homogenate supernatants (at 100,000 xg for 60 min at 4°C) were mixed with 1% triton and re-centrifuged, the obtained supernatants were then incubated with 50 mM sodium pyrophosphate buffer pH 9.5, 2.5 mM NAD⁺(coenzyme) and 10 μM malondialdehyde (substrate). ALDH2 activity was estimated as μmole NADH generated/min/mg protein (Josan et al., 2013). All of the aforementioned assays were normalized using the tissue protein content which was quantified using Bradford (1976) assay (Bradford, 1976).

Total RNA was isolated from the tumor tissues (liver and lung) of LC (untreated and treated) groups using the kit’s standard protocol. Following the quantification of RNA, one-step SYBR green master mix (qPCR kit) with certain primers (Supplementary Table S1) had been used to assess the relative variations in the gene expression using an equation of 2^−ΔΔCt. These genes included Bcl-2-associated X protein (BAX), cyclin-dependent kinase inhibitor 1 (p21), cyclin D, telomerase reverse transcriptase (TERT), vascular endothelial growth factor (VEGF), matrix metalloprotease (MMP)9, ATP binding cassette subfamily G member 2 (ABCG2), CD90, NOTCH1, WNT1, sex-determining region Y-box 9 (Sox9), 4-octamer-binding transcription factor (OCT), and Nanog.

The main liver function parameters were determined in liver homogenate supernatants (1 g liver homogenizing in 9 ml of 150 mM Tris-KCl buffer pH 7.4) after centrifugation at 3,000 g for 15 min at 4 °C (Gaskill et al., 2005). Using colorimetric kits, these parameters (alanine aminotransferase (ALT), aspartate aminotransferase (AST), and albumin) were assessed in both treated LC groups, relative to the untreated LC and healthy N group. A complete blood count (CBC) was also performed on all groups using a hematology analyzer (Mindray, China).

The safety of DF(II) complexes was examined using H&E staining of multiple tissues (liver, lung, spleen, brain, and kidney) of DF(II) NPs- and DF(II)-treated groups versus the untreated healthy N group.

Liver organ of the untreated LC group showed oval pale yellow nodules with increasing its relative weight by 1.7 folds as well as lung weight by 1.6 folds without any increase in spleen weight, compared to tissues of healthy N group (Figures 2B,C). The incidence of metastatic LC was assured by histological investigation at the 6th week of chemical induction, using H&E staining that revealed severe neoplastic alterations (multinucleated giant nuclei with increasing nuclear to cytoplasm ratio) in liver, pneumocytic hyperplasia in lung, and no abnormal variations in spleen section (Figure 2D). At the 9th week, H&E staining demonstrated clear cell hepatocellular carcinoma with hyperchromatic pleomorphic nuclei anaplastic cells in liver section, expanding adenocarcinoma lesion in lung tissue, and lymphoid nodular hyperplasia (abnormal expansion of white pulp) in spleen (Figure 2D). In LC-DF(II) NPs group, there were no nodules in liver and no weight gain in any organs (liver, lung, and spleen) compared to N groups. Meanwhile, liver of LC-DF(II) group had small nodules (mean number 8, compared to 16 nodules in the untreated LC) with a significant increase in liver weight compared to N group but no abnormal increase in lung or spleen weight (Figures 2B,C). H&E staining declared that DF(II) NPs had a higher therapeutic efficacy than DF(II) as evidence by complete area of normal liver, lung, and spleen cells in LC-DF(II) NPs, which is contrary to the presence of neoplastic area in liver and lung sections with expansion of splenic lymphatic nodules in LC-DF(II). This confirms DF(II) NPs completely inhibited LC growth and metastasis to lung or spleen (Figures 2B,C).

Immunohistochemical detection of ki-67⁺ in tumor tissues (liver and lung) and N-cadherin⁺ (metastasis marker) in liver (Figure 3A i,ii) suggested that DF(II) NPs had significantly stronger suppressive potential for these parameters (95.67%, 88.86%, and 96.02%, respectively) than DF(II) complex (36.45%, 64.22%, and 51.19%, respectively). Furthermore, DF(II) NPs normalized the elevation of AFP which was 325 folds in the untreated LC group compared to N group and moderately decreased to 136 folds in LC-DF(II) group (Figure 3B).

Figure 3C demonstrates that DF(II) NPs accumulated mainly in liver tissue (∼70%) followed by lung (7.0%) and spleen (2.32%), while other tissues of LC-DF(II) NPs group recorded very small amounts (≤0.2%) of this nanocomplex, indicating for a highly selective distribution in tumor tissues. More importantly, hepatic and lung uptakes of DF(II) NPs were >3 folds and 2.6 folds, respectively, higher than those of DF(II) as illustrated in Figure 3C. Based on the enhanced permeability and retention of nanoparticles in tumors (Wu, 2021), the results of atomic absorption declared a selective and massive overload of DF(II) NPs, compared to DF(II), in tumor tissues. This accumulation is the main contributor for inducing targeted and efficient ferroptosis (Bekric et al., 2022). The strong trigger of ferroptosis in LC-DF(II) group interprets the perfect eradication of hepatoma cells (Figures 2B,D), which was accompanied by a halt in the elevation of key tumor protein markers, including Ki-67, N-cadherin, and AFP (Figures 3A,B). The marked upregulation of these mentioned markers indicates a high grade of tumor stage, metastatic growth and poor survival (Sell, 2008; Shi et al., 2015; Kaszak et al., 2020). Therefore, it is important to monitor them.

These potent anti-LC activities of DF(II) nanocomplex are mainly attributed to its lipid peroxidation generation-dependent pro-oxidant activity as indicated by elevation of ROS and lipid peroxidation in tumor tissues (liver and lung) by 2.8–6.6 folds and 4.3–6.1 folds, respectively, relative to the untreated LC group (Figures 4A,B). Meanwhile, DF(II) increased ROS and lipid peroxidation contents by only 1.06–2.6 folds and 1.18–1.53 folds, respectively (Figures 4A,B). Moreover, DF(II) NPs suppressed Nrf2 transcriptional activity (2.09–4.3 folds), decreased GSH level (2.10–6.94 folds), and inhibited GPX4 and ALDH2 activities (by > 45.02% and >46.50%) more effectively than DF(II) complex (1.06–2.58 folds, 1.18–1.53 folds, ∼1 fold, 1.096–1.26 folds, 2.97–3.04 folds, and 2.78–3.35 folds, respectively) as shown in Figure 4 C-F. These anti-ferroptotic parameters (Nrf2, GSH, GPX4, and ALDH2) were inactivated by the thiol-affinity of DE forming disulfides resulting in preventing the transcriptional activity (promotor binding) of Nrf2, irreversible oxidation of GSH, and inhibiting GPX4 and ALDH activities (Brar et al., 2004; Abu-Serie & Eltarahony, 2021; Nie et al., 2022). These latter enzymes play key roles in the detoxification of lipid peroxide aldehydes (Bekric et al., 2022; Gao et al., 2022). Recent studies have illustrated that ALDH2 protects against these reactive aldehydes (malondialdehyde and 4-hydroxy-2-nonenal) by converting them into malonic acid and hydroxy-2-nonenoic acid, respectively (Zhao et al., 2019; Gao et al., 2022). Due to higher biodistribution of DF(II) NPs in tumor tissues, this antioxidant protective system was lower in LC-DF(II) NPs and subsequently, this group showed stronger elevation of ROS and lipid peroxidation than in LC-DF(II) group.

Moreover, DF(II) nanocomplex was more effective than DF(II) in upregulating the expression of proapoptotic genes (BAX and p21) and downregulating the expression of all studied oncogenes and stemness genes in tumor tissues of liver and lung (Figure 5 A,B). In both tumor tissues, DF(II) NPs enhanced apoptotic gene expression by 5.5–72 folds and lowered the expression of cyclin D, TERT, VEGF, and MMP9 by ≥ 2 folds and stemness genes (ABCG2, CD90, NOTCH1, WNT1, Sox9, 4-Oct, and Nanog) by ≥ 2 folds compared to <7 folds, ≤2 folds, and ≤1 fold in the case of DF(II), respectively (Figures 5A,B). A previous study illustrated that bismuth diethyldithiocarbamate enhanced cell cycle arrest and apoptosis in HepG2 cells by activation of p53 and BAX (Ishak et al., 2014). This effective alteration in the gene expression of apoptosis, cell cycle, telomerase, angiogenesis, metastasis, chemoresistance, and stemness (NOTCH1, WNT1, Sox9, 4-Oct, and Nanog) in LC-DF(II) is strongly linked to the high accumulative DF(II)-mediated potent ferroptosis-dependent lipid peroxidation, compared to DF(II). As it was demonstrated, DF(II) NPs can suppress functional parameters (ALDH2, CD90, and stemness genes) of CSCs which are responsible for tumor metastasis and chemoresistance (Xiao et al., 2016; Zhang et al., 2018; Wang et al., 2020).

Regarding liver function indexes (Figure 6A), ALT and AST activities were decreased significantly in liver tissues of the untreated LC group (29.49 U/mg protein and 18.37 U/mg protein, respectively) compared to healthy N group (77.26 U/mg protein and 28.82 U/mg protein, respectively). DF(II) NPs were found to be able to normalize this depletion in liver enzymes (72.96 U/mg protein and 25.69 U/mg protein, respectively). On the other hand, DF(II) improved their activities (49.74 U/mg protein and 20.94 U/mg protein, respectively) but still significantly lowered when compared to N group. There was no abnormal variation in the albumin level between the untreated or treated LC groups and healthy N group (Figure 6A). This insignificant difference between the untreated LC group and N group, in the term of albumin level, could be attribute to the protective role of albumin against oxidative damage by preserving cellular GSH (Carvalho & Verdelho, 2018). A recent study declared an inverse correlation between the risk of different cancer types (including, LC) and albumin level (Yang et al., 2021). The normalization effect of DF(II) NPs on the altered liver enzymes ascertained efficient eradication of hepatoma cells.

In terms of hematological indicators (Table 1), the untreated LC groups showed a significant decrease in lymphocyte (Lymph) percentage with increasing monocyte (Mid) and granulocyte (Gran) percentages and no abnormal changes in other parameters relative to healthy N group. These abnormal variations were only normalized in LC-DF(II) NPs group compared to N group and there was no significant improvement in these leukocyte parameters in LC-DF(II) group (Table 1). Lymphopenia is reported in metastatic tumors, particularly non-Hodgkin’s lymphoma, LC, and pancreatic cancer, which suppress the immune response by a variety of mechanisms, including the expression of proapoptotic ligands (Ménétrier-Caux et al., 2019). Therefore, the anti-tumor-dependent ferroptotic effect of DF(II) NPs halted the immunosuppressive potential of metastatic hepatoma cells.