Copper chelation selectively kills colon cancer cells through redox cycling and generation of reactive oxygen species

Human colorectal cancer cells, SW480, HT-29 and LOVO were kindly provided by the American Type Culture Collection (ATCC). Cells were cultured in RPMI 1640 (Sigma-Aldrich, UK) with 20mM HEPES and L-Glutamine at 37°C in a humidified atmosphere of 5% CO2 and 95% air. Media was supplemented with 1% Penicillin-Streptomycin (100 U/ml) and 10% heat-inactivated FBS (Sigma-Aldrich, Germany). Unless otherwise mentioned, cells were seeded at 1.2 ×105 cells/ml and treated with TPEN (Sigma-Aldrich) at 50% confluence. TPEN was prepared in DMSO and the final DMSO concentration used on cells <0.3%.

Human HCT116 p53^+/+ colon cancer cells were cultured as previously described [23]. Cell viability was measured using the MTT-based Cell Titer 96 non-radioactive cell proliferation kit (Promega Corp, Madison, Wisconsin, USA). Cell cycle analyses were performed on propidium iodide stained cells using flow cytometry (Becton Dickinson, Research Triangle, NC). The TUNEL assay used the In Situ Cell Death Detection Kit according to the manufacture instructions (Roche Diagnostics Corporation, Mannheim, Germany). For Annexin V staining cells were incubated in Annexin-V-Fluos labeling solution [20 μl Annexin reagent and 20 μl PI (50 μg/ml) in 1000 μl incubation buffer pH 7.4 (10 mM Hepes/NaOH, 140 mM NaCl, 5 mM CaCl₂), then analyzed by flow cytometry. Caspase 3, 8 and 9 activities were assessed using Colorimetric Assay kits according to manufacturer insutructions (R & D Systems-BF4100). Primary antibody used for Western blots: XIAP #2042S; Caspase 3 #9665S; Caspase 9 #9502S; Bax #2772; PARP #9542S, from Cell Signaling. Cytochrome C sc-13560 from Santa-Cruz and GAPDH #5476 from Abnova.

Cells were treated with TPEN for 10, 20, 30 and 45 min. In experiments which involved addition of the antioxidant N-acetyl-L-cysteine (NAC), cells were treated with 5 mM NAC for 2 h before TPEN after which 10 μM of the CM-H2DCFDA dye was added for 20 min. Cells were washed, harvested by centrifugation and the pellet washed and re-suspended in 500 μl PBS followed by flow cytometery.

Cells were washed, pelleted and incubated in 500 μl of rhodamine buffer [5 μm rhodamine 123, 130 mM NaCl, 5 mM KCl, 1 mM Na₂HPO₄, 1 mM CaCl₂, 1 mM CaCl₂, 1 mM MgCl₂, and 25 mM Hepes (pH7.4)] for 30 min at 37°C, then analyzed by flow cytometry.

10⁶ HCT116 and NCM460 cells were collected in 8 ml HNO₃ 65% + 2 ml H₂O₂ 30% and digested in a closed vessel microwave (Milestone ETHOS PLUS with HPR-1000/10S high pressure rotor). Cell lysates were measured for ion concentration against standard solutions prepared for Copper, Zinc and Iron in deionized water in an atomic absorption spectrophotometer (Furnace).

EPR spin-trapping experiments were carried out with a JEOL-RE1X spectrometer (Kyoto, Japan). Spectrometer settings were as follows: field center, 335.094 mT; microwave power, 10 mW; sweep time, 2.0 min; time constant, 0.3 s; and modulation width, 0.2 mT). UV/VIS spectra were recorded with Helios Alpha spectrophotometer (Thermo Fisher Scientific, Inc.; Pittsburgh, PA). All measurements were performed at room temperature.

NOD/SCID female mice, (6–8 weeks, ~20 g) (Charles River Laboratories, France) were divided into two groups of 8 mice each and maintained in Maxi-miser hepafiltered facility. 2-3×10⁶ HCT-116 tumor cells in 100 μl of 0.9% NaCl were inoculated subcutaneously in the flank. An average age (6–8 weeks) and body weight of mice (18-22 g) were used for the experiments. Prior to manipulations, mice were anaesthetized with isoflurane (Forane®, Abbott) by inhalation. On day 7 post inoculation mice received i.p. injections of either saline (control) or 20 mg/kg TPEN every other day for 28 days. Tumor measurements were performed 3 times per week using a sterile Vernier caliper. Tumor volume was calculated by the formula: Volume = π/6 (length × width × height). A 5-10% loss of body weight was detected in mice receiving TPEN treatment in comparison to the control but they otherwise looked healthy and tolerated the drug treatment well. The use of laboratory animals was in accordance with the guidelines of the Institutional Animal Care and Use Committee (IACUC). The IACUC committee at the American University of Beirut where the animal studies were conducted reviewed and approved the studies described.

Tissue sections (4 μm) were stained with and anti-Ki-67 antibody (Santa-Cruz, US), for 60 min followed by secondary and tertiary antibodies and incubated with the chromogen (Zymed, US) before counterstaining with Hematoxylin. To assess the extent of total cell death, tissue sections were stained by using the terminal deoxyribonucleotidyl transferase-mediated dUTP-nick-end labeling (TUNEL) assay, according to manufacturer’s instructions (in situ cell death detection kit, fluorescein; Roche) and counterstained with PI. Slides were analyzed under LSCM fluorescent confocal microscope (LSM 410, Zeiss, Germany).

Data are expressed as the mean ± standard deviation, and statistical significance between different groups was determined using a two-tailed Student’s t-test. Statistical significance was defined as a *p < 0.05 and **p < 0.01.

Treatment of HCT116 colon cancer cells with 5 μM TPEN for 24 h results in a noticeable decrease in cell density (Figure 1A). In contrast, non-cancerous colon (NCM460) or intestinal (FHS74Int) cell lines (see methods for cell lines description) were resistant to TPEN at these concentrations (Figure 1A). A dose response of TPEN toxicity in the two non-cancerous (NCM460 and FHS74Int) and in four independent human colorectal cancer cell lines (HCT116, SW 480, HT-29, and LOVO), consistently revealed a significantly higher sensitivity of the cancerous cells as compared to their non-cancerous counterpart. The IC₅₀ for the non-cancerous cells was 8.3 μM and 13μM for the NCM460 and FHS74Int respectively. In contrast the IC₅₀ for the cancerous cells were 2.7; 4.0; 4.3 and 4.5 μM for the HCT116; SW480; HT-29 and LOVO cells respectively. Importantly, selective killing of cancerous cells (15.3 ± 0.4%; 25.07 + 6%; 10.82 + 1.18%; 33.35+ 5.22% survival for HCT, SW, LoVo, HT respectively) could be achieved at 5 μM TPEN for 24 h, concentrations at which 96.9 ± 7.52% and 95.3 ± 1.7% of the NCM460 and FHS74Int cell lines remained viable (Figure 1B). Flow cytometry analysis in HCT116 cells showed that TPEN toxicity is time-dependent as indicated by the gradual enrichment of the pre-G₁ percentage in the cell population (Figure 1C). No obvious cell cycle arrest was associated with the TPEN treatment (Figure 1C), indicating that the primary effect of TPEN on HCT116 cells is to induce cell death.

The observed increases in the PreG₁ population are associated with higher levels of apoptosis, as indicated by TUNEL and Annexin V assays in TPEN treated cells in a dose-dependent fashion (Figure 1D). An early hallmark of most apoptotic cell death is increased mitochondrial membrane permeability. TPEN induced permeabilization of the mitochondrial membrane, as indicated by changes in mitochondrial membrane potential (ΔΨ_m) (Figure 1E). Interestingly, this effect may be associated with TPEN’s ability to “extract” essential metals, particularly iron and copper from mitochondrial electron carriers, thus facilitating the loss of membrane potential and the deviation of electron flow to molecular oxygen O₂ to generate superoxide and H₂O₂. Treatment with TPEN induced disruption of the mitochondrial membrane potential in a dose-dependent fashion, as detected by the reduced accumulation of the fluorescent dye rhodamine (Figure 1E). Consistently, disruption of the mitochondrial membrane is associated with the release of cytochrome c from the mitochondria and its accumulation in the cytosolic fraction (Figure 1F).

These results argue that TPEN induces apoptosis in HCT116 cells through the intrinsic pathway. In agreement with this conclusion, TPEN modulated key markers of apoptosis resulting in cleavage of caspases 3 and 9 into their active products (Figure 1G), cleavage of PARP, and accumulation of the pro-apoptotic BAX, which is typically observed upstream of mitochondrial dysfunction (Figure 1H). TPEN killing of cancer cells was reported to be associated with degradation of the X-linked inhibitor of apoptosis (XIAP) [15, 24]. Consistently, treating HCT116 cells with TPEN results in XIAP degradation (Figure 2A, XIAP). XIAP degradation is blocked by incubations with zinc, showing that it is dependent on TPEN’s metal chelation properties (Figure 2A). However, inhibition of caspases with the pan-caspase inhibitor z-VAD was ineffective in blocking XIAP degradation (Figure 2C), arguing that TPEN-dependent XIAP degradation occurs upstream of caspase activation. These data show that TPEN induces apoptosis with all the associated molecular hallmarks.

The ability of TPEN to induce death of HCT116 colon cancer cells was dependent on its metal chelation properties, as it was reversed when TPEN was added in the presence of ZnSO₄ (Figure 2B, Zinc). TPEN is a metal chelator with high affinity for zinc, copper and iron. Addition of exogenous Zn²⁺ is expected to associate with and saturate TPEN, preventing it from chelating other metals in the cell. ZnSO₄ addition effectively abrogated the toxic effects of TPEN as indicated by the cell viability assay (Figure 2B, Zinc), TUNEL and annexin staining (Figure 2C, Zinc) and PARP, caspase 3 and XIAP cleavage (Figure 2A, Zinc). In contrast, inhibition of caspases using the pan-caspase inhibitor z-VAD-fmk resulted in significant, yet only partial reversal of TPEN toxicity (Figure 2B and 2C). z-VAD treatment effectively blocked caspase 3 and PARP cleavage (Figure 2A), this however was not sufficient to fully reverse TPEN-dependent cell killing (Figure 2B and 2C, z-VAD). These results suggest that the toxic effects of TPEN are mediated only partially through the induction of apoptosis, and potentially invoke other death pathways such as necrosis and autophagy.

To investigate mechanisms involved in TPEN-dependent cell killing, we tested the involvement of reactive oxygen species (ROS) production using the DCFDA assay, since TPEN has been previously implicated in ROS production [25]. TPEN treatment caused a significant increase in intracellular ROS as early as 10 min after treatment, which gradually decreased over the next 30 min (Figure 3A). This ROS increase in HCT116 was significantly higher than that observed following the addition of 250 μM of H₂O₂ (Figure 3A). In contrast, TPEN-induced ROS production in the NCM460 normal colon cells was significantly smaller (Figure 3A), and decayed back to baseline with a faster time course than in HCT116 cells (Figure 3B). Similar to the data obtained in HCT116 cells, TPEN treatment of two other colon cancer cell lines, SW480 and LOVO, also resulted in the production of significantly higher levels of ROS as compared to the normal NCM colon cells (Additional file 1: Figure S1). These data show that TPEN induces a 6-7fold higher ROS in colon cancer cells as compared to the normal NCM460 cells, and that this increased ROS lasts for tens of minutes longer in HCT116 cells (Figure 3A, 3B and Additional file 1: Figure S1). Hence there is good correlation between ROS induction and TPEN toxicity in these two cell lines.

To assess whether ROS generation is involved in mediating TPEN-induced cell death, we pretreated HCT116 cells with antioxidants, including NAC and vitamin E, after which cell viability and apoptosis were assessed. As expected, the general ROS scavenger NAC and vitamin E significantly decreased H₂O₂- and TPEN-induced ROS generation (Additional file 2: Figure S2). Antioxidants effectively prevented TPEN toxicity as measured by cell viability (Figure 3C), TUNEL and annexin assays (Figure 3D), and reversed caspase 3 and 9 activation (Additional file 3: Figure S3). This argues that TPEN-mediated cell killing requires ROS production. Furthermore, the protective effects of antioxidants are also compatible with the idea of transition metal catalyzed oxidative stress. However, the observed incomplete protection by two different antioxidants – NAC and Vit E – may indicate contribution of non-redox driven pathways, such as Zn-dependent effects, given previous reports that argued for a role of Zn-homeostasis in TPEN toxicity. Nonetheless, the significant reversal of the TPEN toxicity using two different anti-oxidants strongly argues for an essential role for ROS production in TPEN toxicity.

We were then interested in defining the mechanism by which TPEN induces ROS generation and cell killing. In the presence of ascorbic acid, TPEN has been shown to mobilize iron from ferritin, presumably via formation of a [TPEN-Fe^(II)] complex. This reaction may have toxicological consequences, as [TPEN-Fe^(II)] can participate in Fenton-like reactions with concomitant generation of hydroxyl radical and/or high-valent iron ^(IV)-oxo species. In addition, TPEN was suggested to deplete intracellular copper in several cell types [11, 26, 27]. Hence, we carried out experiments to verify whether ascorbic acid can reduce [TPEN-Fe^(III)] to [TPEN-Fe^(II)], and [TPEN-Cu^(II)] to [TPEN-Cu^(I)]. This is important in the context of the ROS observed in response to TPEN treatment because the reduced forms of these complexes may have the potential to generate hydroxyl radical in the presence of H₂O₂.Ascorbic acid reduces [TPEN-Fe(III)] at a relatively low rate (Figure 4A, closed circles), thus setting the stage for induction of a futile redox cycle whereby [TPEN-Fe(II)] autoxidizes to [TPEN-Fe(III)] with concomitant generation of superoxide anion radical. Incubation of ascorbic acid with [TPEN-Cu(II)], however, did not result in changes in the electronic spectrum of the complex (Figure 4B). This suggests that either ascorbate does not possess the required potential to reduce [TPEN-Cu(II)] to [TPEN-Cu(I)], or the latter rapidly autoxidized with generation of superoxide anion radical. To verify the latter hypothesis, we assessed the ability of [TPEN-Cu(II)] to generate hydroxyl radical in a reaction system consisting of H2O2, ascorbic acid, dimethylsulfoxide (CH3SOCH3; DMSO) and N-tert-butyl-alha-phenylnitrone (PBN) as a spin trapping agent. In the complete reaction system, the initial Electron Paramagnetic Resonance.

(EPR) doublet of the semidehydroascorbyl radical (ascorbate + radicals → semidehydroascorbyl radical) was gradually substituted by the characteristic 6-line EPR spectrum of PBN/^.CH₃ (Figure 4C; CH₃SOCH₃ + HO^. → ^.CH₃ + CH₃S (O) OH; ^.CH₃ + PBN → PBN/^.CH₃ ^.). The formation of PBN/^.CH₃ was inhibited by depletion of ascorbic acid, as assessed by the disappearance of the semidehydroascorbyl radical. No PBN/^.CH₃ formation was observed if either H₂O₂ or ascorbic acid was omitted from the reaction system. Under these experimental conditions, the rate of PBN/^.CH₃ formation by [TPEN-Fe^(III)] was 4 times slower than that observed with [TPEN-Cu ^(II)] (Figure 4D). The results obtained indicate that both [TPEN-Fe^(II)] and [TPEN-Cu ^(I)] have the potential to generate HO^. in biological systems ([TPEN-Me⁽ⁿ⁺⁾] + H₂O₂ → [TPEN-Me⁽ⁿ⁺¹⁾] + HO^. + HO^-). These observations extend the reaction mechanism of the breakdown of H₂O₂ by ([TPEN-Fe⁽ⁿ⁺⁾], which includes the formation of high-valent oxo-species that can activate C-H bonds and epoxidate > C = C < bonds [28‐30]. Hence the TPEN-Fe and TPEN-Cu complex engage in redox cycling resulting in the generation of ROS. As such chelation of cellular copper and/or iron by TPEN has the potential to produce ROS leading to cellular toxicity.

Given the potential of TPEN-Cu ^(II) and TPEN-Fe^(II) to redox-cycle in the presence of H₂O₂ and a reductant (eg, ascorbate) to yield hydroxyl radicals, the ability of these TPEN-metal complexes to produce cellular damage is likely to be proportional to the amount of chelatable metals in cells. Tumor tissues from multiple cancers have been reported to have elevated copper content [22]. This is also the case for HCT116 cells as shown by atomic absorption spectroscopy studies, where both copper (enriched ~7-folds) and zinc (enriched ~5.5-folds) accumulate to significantly higher levels as compared to NCM460 (Figure 4E). In contrast, iron content was similar between the two cell types (Figure 4E). Similar accumulation of copper and zinc but not iron were observed in two additional colon cancer cell lines, LoVo and SW480, where the measured levels for Cu were 120.25 and 99.56 μg/L; Zn 20.34 and 15.26 μg/L; Fe 44.24 and 46.25 μg/L in LoVo and SW480 cells respectively. Given that zinc is not redox active and that the levels of iron are similar in both cell types, the mechanisms of action of TPEN is likely to involve formation of a TPEN-Cu complex that engages in redox cycling to generate ROS and cause cellular damage. Interestingly the fold enrichment of copper in HCT116 cells (Figure 4E) mirrors the fold increase in redox-cycling activity revealed by DCFH oxidation measured in these cells as compared to NCM460 following TPEN treatment (Figure 3A).

The above data support a model where TPEN induces selective death of cancer cells through its ability to chelate copper away from cellular proteins, resulting in the formation of a TPEN-copper complex able to engage in redox cycling thus generating ROS and inducing cell death. Interestingly, the TPEN-Cu complex is predicted to be impermeant to the cell membrane, given the positive charges contributed by the copper ion, thus trapping the complex intracellularly and increasing the levels of ROS produced. This model makes two testable predictions. First, it predicts that TPEN is capable of stripping a significant proportion of copper away from cellular proteins. To determine whether this is the case, we measured the fraction of total cellular copper that is chelatable by TPEN under our experimental conditions. As shown in Figure 4F, ~50% of the total cellular copper can be chelated by TPEN following incubation with 5μM TPEN for 24 hrs.

The second prediction from the proposed model is that cellular copper is essential for TPEN toxicity. Indeed if TPEN induces cell death through redox cycling after the formation of a TPEN-Cu complex, then reducing the enriched total cellular copper in cancerous HCT116 cells would be expected to reduce TPEN toxicity. To test whether this is the case we incubated HCT116 with the membrane permeant copper specific chelator neocuproine [31]. We chose neocuproine because of its specificity for copper and because the neocuproine-Cu complex in unlikely to be redox active [32]. Therefore, preincubation of HCT116 cells with neocuproine is expected to chelate cellular copper thus preventing TPEN from binding it. In the absence of the formation of TPEN-Cu complexes TPEN treatment should not produce ROS. Neocuproine by itself mildly reduces cellular viability (Figure 5A, Neo), probably due to the requirement for copper for normal cellular homeostasis. Importantly, incubating cells with neocuproine does not generate ROS (Figure 5B, Neo), showing that indeed the Neo-Cu complex is not redox active. However, pre-incubating HCT116 cells with neocuproine significantly reverses the toxic effect of TPEN (Figure 5A, Neo + TPEN), and this reduction in toxicity is coupled to a significant reduction in the levels of ROS generated (Figure 5B, Neo + TPEN5).

To test whether neocuproine reverses TPEN toxicity by chelating intracellular copper, we incubated cells with a membrane impermeant copper specific chelator, bathocuproine [33]. In contrast to neocuproine, bathocuproine did not reverse the cellular toxicity due to TPEN treatment (Figure 5C). Collectively these data show that TPEN toxicity requires cellular copper. When intracellular copper is chelated with neocuproine, a copper chelator that is effective at chelating copper but does not allow the TPEN-Cu complex to engage in redox reactions, TPEN becomes ineffective at producing ROS and killing colon cancer cells. Furthermore, as shown in Figure 2A with zinc, copper was also effective at reversing the toxic effects of TPEN and its ability to generate ROS in a dose dependent fashion (Figure 5A and B), showing that TPEN toxicity is dependent on its metal chelation properties.

The ability of TPEN to selectively kill cancer cells raises the attractive possibility that it could have anti-tumor potential in vivo. To test whether this is the case we used a mouse xenograft colon cancer model. HCT116 colon cancer cells (2-3×10⁶), were subcutaneously inoculated into the flank of NOD/SCID mice, and showed efficient engraftment as reflected by the development of malignant palpable tumor mass within 7–9 days post-inoculation. To investigate the therapeutic efficacy of TPEN, xenografted mice were injected intraperitoneally (i.p.) at 7 days post-inoculation with TPEN (20 mg/kg) every other day for a 22 days therapy span. Tumor size was monitored in mice at different time points starting at day 7 post inoculation for both groups. In the TPEN treated group tumor volume was significantly reduced throughout the time course of tumor development with a volume of 11.6 ± 0.37 cm³ in the control group versus 3.75 ± 0.142 cm³ in the TPEN-treated group at day 22 (Figure 6A). In our hands TPEN was well tolerated at 15–20 mg/kg, but not at 30 mg/kg, consistent with previous reports [34].

Histological H & E staining of tumor sections from untreated mice showed malignant cell infiltration and retention in the tissue (Figure 6B). However, TPEN-treated group showed a significant decrease of malignant cell nuclei staining relative to the control mice, consistent with the tumor growth profile (Figure 6A and B). In addition, the expression of Ki-67, a nuclear protein proliferative marker, was reduced in TPEN-tissue section as depicted by immunostaining (Figure 6B). Furthermore, in vivo TPEN treatment induced apoptosis in tumor tissue as reported by TUNEL staining (Figure 6C). Collectively these data show that TPEN is effective at inhibiting the growth of colon cancer cells in vivo, arguing that metal chelation is a promising intervention for colon cancer treatment.