Molecular features of the cytotoxicity of an NHE inhibitor: Evidence of mitochondrial alterations, ROS overproduction and DNA damage

Colon carcinoma HCT-116 cells (from Dr. C.R. Boland [13]) were cultured in D-MEM supplemented with 10 % foetal calf serum (FCS), 4 mM glutamine, 2 mM Na pyruvate, 100 U/ml penicillin and 0.1 mg/ml streptomycin. Colon adenocarcinoma HT-29 cells (obtained from Dr. R. Supino [14]) were grown in McCoy medium supplemented as for D-MEM. Reagents were purchased from Thermo Fisher (Waltham, MA, USA) and Euroclone (Milano, Italy). Cells were grown as monolayer at 37 °C in humidified atmosphere containing 5 % CO₂ in 75 cm² flasks (Corning, Amsterdam, The Netherlands). At confluence, cells were trypsinised with trypsin-EDTA (stock solution 10X, Sigma-Aldrich, Saint Louis, MO, USA) diluted in PBS (Thermo Fisher).

Cell lines were treated with HMA (Sigma-Aldrich, stock solution 80 mM in DMSO) at concentrations ranging from 5 μM to 40 μM (accordingly to [10]) for the time periods indicated in the text. In some experiments, cells were treated with 15 mM NAC (N-acetyl cysteine, Sigma-Aldrich, stock solution 100 mM); 20 μM zVAD.fmk (Bachem, Heidelberg, Germany, stock solution 20 mM); 50 μM Necrostatin-1 (Biomol, Hamburg, Germany, stock solution 38.56 mM); 10 μM Olaparib (SelleckChem, Houston, TX, USA, stock solution 100 mM); 400 μM α-Tocopherol (Sigma-Aldrich, stock solution 22.36 mM). Parallel samples were incubated with 0.1 % DMSO, i.e. the same final concentration used with the highest drug concentration.

Cells (either unfixed or being fixed with 4 % paraformaldehyde (PFA) in PBS) were analysed in bright field microscopy.

For transmission electron microscopy (TEM), cells were processed as follows. Some cells were fixed with 2.5 % (v/v) glutaraldehyde and 2 % (v/v) PFA in 0.1 M phosphate buffer; pH 7.4, at 4 °C for 1 h, washed and incubated with 3,3’ diaminobenzidine (DAB, Sigma-Aldrich) (2 mg/ml in 50 mM Tris-HCl; pH 7.6) under simultaneous irradiation with two 8 W Osram Blacklite 350 lamps for 2 h at room temperature (spectral emission range between 430 and 470 nm, thus being suitable for FITC excitation). The cells were then post-fixed with 1 % osmium tetroxide (Electron Microscopy Sciences, Hatfield, PA, USA) and 1.5 % potassium ferrocyanide (Sigma-Aldrich) at room temperature for 1 h, dehydrated with acetone and embedded in Epon (Agar Scientific, Assing, Monterotondo, Italy). Parallel samples, after aldehyde fixation, were incubated with DAB under light irradiation, dehydrated with ethanol and embedded in LR White resin. As negative controls, some samples were processed as described above but omitting both DAB incubation and exposure to the excitation light. Ultrathin sections were weakly stained with uranyl acetate (Agar Scientific). The samples were finally examined and photographed with a Zeiss EM 900 electron microscope at 80 KV (Carl Zeiss, Jena, Germany). The micrographs were then developed and digitalised [15].

Cells were seeded at the density of 2x10⁴/well in 24-well plates. After 24 h, cells were treated with the selected drug. At the end of the treatment, the cells were rinsed with PBS and trypsinised, washed in binding buffer (0.1 mM Hepes, 2.5 mM CaCl₂, 140 mM NaCl; pH 7.4) and incubated for 15 min with Annexin V-FITC (Milteny Biotech, Bergisch Gladbach, Germany, 3 % in binding buffer). Samples were then washed again in binding buffer, stained with PI (propidium iodide, Sigma-Aldrich, diluted 1:1000 in binding buffer) and immediately analysed by flow cytometry (FACSCanto II, BD Biosciences, Heidelberg, Germany).

The expression level of a panel of proteins in total extracts was analysed by Western blot according to procedures routinely used in the laboratory [10, 16] using the primary antibodies listed in Table 1. HRP conjugated secondary antibodies were from Jackson Immuno-Research (Suffolk, UK). Three independent experiments were carried out. Band quantification was performed using ImageJ software (https://​imagej.​nih.​gov/​ij/​).

Cells were seeded on coverslips (16-mm diameter) in 12-well plates, at the density of 5x10⁴ cells/ml. After 24 h, cells were treated with the appropriate drug concentration for increasing times (up to 24 h). The following antigens were probed:

Cells were seeded in 24-well plates at the density of 2x10⁴/well. After 24 h, cells were treated with the appropriate drug for the selected time; thereafter, the fluorogenic dye DCFDA (dichlorofluorescein diacetate, Invitrogen, Molecular Probes, 10 μg/ml) was added to the medium for 30 min. After diffusion into the cell, DCFDA is deacetylated by cellular esterases to a non-fluorescent compound, which is later oxidized by ROS into 2’,7’-dichlorofluorescein (DCF). DCF is a highly fluorescent compound, which can be detected by fluorescence spectroscopy with maximum excitation and emission spectra of 495 nm and 529 nm, respectively. Finally, cells were trypsinised, resuspended in PBS and analysed by flow cytometry (FACSCanto II, BD Biosciences). Each experimental point was conducted in triplicate.

To determine the mitochondrial membrane potential, cells were seeded in 24-well plates at the density of 2x10⁴/well, treated with the proper drug and then incubated with 50 nM TMRM (tetramethylrodamine methylester) (Invitrogen, Molecular Probes) for 10 min at 37 °C. Thereafter, samples were trypsinised, resuspended in 100 μl of PBS and immediately analysed by flow cytometry (FACSCanto II, BD Biosciences).

For transient knockdown, cells seeded in 24-well plates (for death analysis) or in 6-well plates (for western blot analysis) at the density of 7.5x10⁵/well cells were reversely transfected with 5 nM Silencer select control siRNA (4390844) or targeting siRNA (s21741 and s21741 for RIPK3; s20650, s20651 and s20652 for Atg7; s16537 and s16538 for Beclin-1), using Lipofectamine RNAiMAX reagent and Opti-MEM medium according to manufacturers’ protocols. All reagents were purchased from Thermo Fisher.

A preliminary survey of HMA effects on a panel of human cancer cell lines proved that HMA exerted a cytotoxic effect on three colon cancer cell lines, with an IC₅₀ value of 22.30 μM for SW613-B3, 23.47 μM for HCT-116 and 32.19 μM for HT-29 cells [10]. Now, we aimed at depicting the molecular events leading to HMA cytotoxicity. First, we exploited the fluorescent properties of HMA [7] using the diaminobenzidine (DAB) photo-oxidation technique, which is based on the formation of an electron-dense osmiophilic product that precipitates in close proximity to the fluorophore, thereby allowing its ultrastructural detection [15, 19]. Electron microscopy analysis of HMA-treated HCT-116 colon carcinoma cells revealed electron-dense black spots within the cytoplasm, enclosed within vesicles (red*), which are present only as an effect of HMA, given that in untreated cells no evidence of roundish deposit of electron-dense products was found (Fig. 1a). In addition, in HMA-treated samples we detected some vesicles containing intracellular debris that can be likely associated to residual cisternae and resemble multilamellar bodies (red arrowhead) [20]. In fact, vacuole-like structures and swelling of intracellular structures were detected also by optical microscopy in the same conditions, exclusively in HMA-treated cells (Fig. 1b). On the whole, these observations suggest that HMA induces some mechanical stress due to the massive presence of vacuoles.

After verifying that HMA had entered the cell and remained in the cytoplasmic compartment, where it promoted morphological alterations, we addressed its possible impact on the cytoplasmic organelles that play a master role in controlling cell metabolism, i.e. mitochondria. Their intracellular distribution was monitored by indirect immunofluorescence experiments using an antibody specific to the mitochondrial chaperone mtHSP70 (red fluorescence). We observed that mitochondria appeared to be rearranged in structure and localisation in HMA-treated HCT-116 cells: in untreated samples, mitochondria were distributed throughout the cytoplasm, while in HMA-treated cells they were detected as clustered outside the nuclear membrane (Fig. 1c).

As mitochondrial functions are strictly correlated with ROS production, we have measured the ROS level through a cytofluorimetric assay based on the use of the fluorogenic dye DCFDA that measures hydroxyl, peroxyl and other ROS within the cell; once diffused into the cell, it is deacetylated to a non-fluorescent compound, which is later oxidized by ROS into the highly fluorescent compound DCF [21]. In fact, as illustrated in Fig. 2a, in HMA-treated HCT-116 cells the quantification of DCF fluorescence intensity, which is proportional to ROS amount, showed a net time-dependent increase. When cells were cotreated for increasing times (up to 24 h) with well-known antioxidants, i.e. NAC (N-acetyl-L-cysteine) and α-Tocopherol (Toc), ROS levels were drastically reduced (Fig. 2a), indicating that both scavengers were effective in counteracting ROS production triggered by HMA.

To monitor mitochondrial transmembrane potential, we used TMRM, a lipophilic cation that normally penetrates the mitochondrial lipid bilayer and accumulates in the mitochondrial membrane emitting a fluorescence peak at 574 nm measurable by flow cytometry while it does not enter altered mitochondria, thus lowering the detectable intramitochondrial fluorescence. A sharp peak of fluorescence corresponds to cells with healthy mitochondria incorporating and accumulating the dye, whereas cells presenting depolarized (low fluorescence emission) or hyperpolarized (high fluorescence emission) mitochondria can be identified at the left and right side of the fluorescence peak of untreated cells, respectively. For each sample, the number of HCT-116 cells included in each class was recorded. No high fluorescence emission/hyperpolarization was observed; the ratio between cells with low (HMA-treated samples) and normal (untreated samples) fluorescence emission was calculated and reported as fold increase. As shown in Fig. 2b, HMA induced a net depolarization of the mitochondrial membrane yet after 4 h of treatment (8.33 folds), which increased even more at 8 h (11.5 folds) while a prolonged incubation allowed the cells to partially buffer mitochondrial membrane depolarization (8.44 and 6.44 fold at 16 h and 24 h, respectively). Similar results were obtained with the HT-29 colon cancer cell line (data not shown). NAC protected the cells from the HMA-induced depolarization, while no effect was observed with α-Tocopherol (Fig. 2b). This discrepancy could be due to the fact that the α-Tocopherol is a lipophilic cation that enters mitochondrial membranes only if mitochondria have an intact membrane potential, while the mechanism of action of the water-soluble NAC implies a rapid reaction with highly oxidising radicals.

A high level of ROS is very dangerous for the cell, being a source of oxidation of DNA and organelles. We monitored the production of 7,8-dihydro-8-oxoguanine (8-oxoG), which is the most frequent oxidation product in both DNA and RNA and possibly contributes to various inflammatory processes and aging-related diseases [22]. HCT-116 cells were analysed by conventional in situ immunolabeling with a monoclonal antibody against 8-oxoG [23]. As shown in Fig. 2c, untreated cells were negative for the presence of 8-oxoG, while in all the cells treated for 24 h with 20 μM HMA, brilliant green fluorescent foci corresponding to the formation of 8-oxoG were clearly visible, confirming the presence of oxidised bases previously observed by the comet assay in HMA-treated cancer cells, thus supporting the postulated correlation between ROS production and base oxidation [10]. In parallel samples treated with NAC in combination with HMA, few foci were still detectable, possibly due to a low residual ROS amount (Fig. 2c).

The comet assay previously applied to HMA-treated cells showed a net increase of single- and double-strand breaks (SSBs and DSBs) [10]; here, we monitored the γ-H2AX form of the H2AX histone that is phosphorylated when DSBs are present in DNA [24]. In fact, as shown in Fig. 2d, a high fraction of HMA-treated cells (57.96 % ± 3.62), showed many red fluorescent nuclei (not visible in untreated cells), as expected in γ-H2AX positive cells. Together, these data support the notion that HMA was able to affect DNA integrity, possibly via ROS production.

The presence of DNA damage, a high amount of ROS together with compromised mitochondria, as well as alterations in cell morphology after HMA treatment, could have an impact on cell viability. We stained cells with PI, which does not enter living cells, while it penetrates dying/dead cells, and analysed them by flow cytometry. HCT-116 cells treated with increasing concentrations of HMA (10-40 μM) for 24 h revealed a highly significant (P < 0.001) dose-dependent increase in the amount of PI⁺/dead cells, reaching about 50 % at 35 μM and remaining significant at 40 μM HMA (P < 0.001), even if the net number of PI⁺ cells declined to about 40 % (Fig. 3a). No effect on cell viability was recorded in samples incubated with the drug solvent DMSO (not shown).

How do HMA-treated cells die? We previously reported that HMA was unable to trigger the final steps of canonical apoptosis even if it promotes the activation of the initiator caspase 8 and 9 [10]. However, given that caspase 8 could be involved in a cross talk between apoptosis and other forms of death [25], here we used the pan-caspase inhibitor zVAD.fmk in combination with HMA. Cells treated with 20 μM zVAD.fmk alone showed a low amount of PI-permeable cells, similar to untreated samples. PI-stained cells after a 24-h treatment with 30 μM and 40 μM of HMA in the presence of zVAD.fmk (20 μM) revealed a significant (P < 0.001) decrease in the number of PI positive/dead cells compared to HMA treatment alone (Fig. 3b), accounting for 38.61 % ± 0.75 (30 μM HMA) and 25.42 % ± 4.16 after HMA/zVAD.fmk treatment. Analogously, for the 40 μM HMA concentration, the fraction of PI positive cells decreased from 46.79 % ± 1.76 to 38.38 % ± 5.92, highlighting a positive effect of the caspase inhibitor on cell viability and suggesting that initiator caspase 8 and 9 could be involved in other subroutes of death driven by HMA.

With this in mind, we investigated necroptosis, a death process where a role for caspase 8 has been described [26]. The possible involvement of necroptosis in the response to HMA was first investigated in HCT-116 cells by using the RIPK1 (receptor-interacting serine/threonine-protein kinase1) inhibitor Necrostatin-1 (NEC, 50 μM), which did not affect cell viability per se (Fig. 3b). When administered together with HMA (30 μM and 40 μM) for 24 h, NEC did not rescue HMA-induced cell death (Fig. 3b), thus suggesting that in HCT-116 cells RIPK1 is not involved in the cellular response to HMA, as already shown in breast cancer cells [9]. To go deeper into the necroptosis issue by addressing the impact of the other key regulator RIPK3, we used the HT-29 cell line, being HCT-116 cells characterised by a low expression of RIPK3 [27].

Western blot analysis of the expression of necroptosis effectors RIPK1 and 3 and MLKL (mixed lineage kinase domain-like) in untreated and HMA-treated HT-29 samples. We observed a modulation in response to the drug treatment, with an increase in RIPK3 and MLKL proteins in HMA-treated samples with respect to controls (1.60 and 1.97 fold, respectively; P < 0.01) (Fig. 3c); however, in this cell line an opposite trend was recorded for RIPK1 (0.60 fold decrease; P < 0.01). As reviewed by Lalaoui et al. [28], the requirement of RIPK1 in necroptosis is not absolute and cells lacking or expressing low levels of RIPK1 (as it is the case of HT-29 cells) undergo necroptosis by spontaneously increase the expression levels of RIPK3 and MLKL, as here observed.

Focusing on RIPK3 in HT-29 cells, we then used a different experimental approach, based on silencing of RIPK3. In HT-29 cells silenced for RIPK3 expression (Fig. 3e), the cell population characterised not only by PI permeability but also by late phosphatidylserine expression detected with Annexin V was decreased upon treatment with HMA, leading to an increase of the number of living cells from about 70 % to about 90 % (Fig. 3d). This suggests that the kinase RIPK3 is required, at least in part, for cell death induction by HMA.

We previously detected some autophagy markers in HMA-treated colon carcinoma SW613-B3 cells [10], as also confirmed in breast cancer cells [9]; now, we addressed whether HMA drives a pro-survival or pro-death role of autophagy. To do this, we silenced Atg7 and Beclin 1 autophagy effectors (Fig. 4b) and then we analysed HCT-116 cell viability by Annexin V/PI staining. Both Atg7 and Beclin-1 silencing caused a decrease in unstained/alive cells of about 40 % upon HMA-treatment (from 79.03 % of control siRNA to 37.27 % for Atg7 and 43.43 % for Beclin-1), as well as enhanced cell death (Fig. 4a) supporting a protective role of autophagy in the presence of HMA-induced stress. However, autophagy was not completely efficient in rescuing cell viability after HMA treatment.

To address this intriguing point, we investigated the last events in the autophagy pathway. The final steps of autophagy are characterised by the sequestration of factors to be discarded through the action of the protein p62, also called sequestosome, which drives the cargo within autophagosomes, and is further degraded. In HMA-treated cells, p62 was detected as several spots (green fluorescence) and clustered both at nuclear and cytoplasmic level independently of the drug concentration (20 μM or 40 μM) (Fig. 4c). In fact, the analysis of the global ubiquitination level using an antibody against ubiquitin, revealed a low red fluorescent signal in untreated samples, while in HMA-treated cells several brilliant regions were visible (both in nuclear and cytoplasmic regions), possibly coincident with p62 distribution (Fig. 4d). Moreover, we analysed the UPR (unfolded protein response) marker GRP78 [29]: in control cells, GRP78 staining appeared as a low diffuse fluorescence, while in HMA-treated cells (20 μM) several specific dots were visible, indicating that GRP78 protein remained accumulated after the drug insult (Fig. 4d). Globally, our results point to a defect in the last phase of autophagy, leading to an impaired clearance of stress-damaged proteins and to an incomplete protection against HMA-induced insults.

The evidence that HMA triggers DNA damage and oxidative stress conditions prompted us to investigate this effect on a cellular emergency reaction, i.e. poly(ADP-ribosylation), which we have already demonstrated to be activated by HMA [8, 10]. We addressed this issue in our experimental conditions by using the PARP inhibitor Olaparib (OLAP) under conditions abolishing PAR synthesis triggered by 20-μM HMA exposure, that are 10 μM for 24 h (Fig. 5a). First, we evaluated cell viability by the metabolic MTT assay, observing once more the inhibitory effect of HMA alone, causing a reduction of cell viability to 70.48 % ± 2.96 and, when combined to OLAP, decreasing to 50.24 % ± 4.74 (Fig. 5b), suggesting a net effect of the PARP inhibitor, so a role of PAR in modulating cell responses to HMA-treatment. Is PAR pro-survival function connected to autophagy? To answer this question, we monitored the autophagy marker LC3, generally overexpressed in HMA-treated cells [9, 10]; in samples co-treated with OLAP, a significant reduction of the LC3 signal with respect to the HMA treatment alone was observed (Fig. 5c), suggesting a possible requirement of PAR in the autophagy machinery [25, 30]. However, given the pleiotropic effects of PAR in cell metabolism, further experiments are required to better understand this issue.