|Home | About | Journals | Submit | Contact Us | Français|
Malignant mesothelioma (MM) is an intractable tumor of the peritoneal and pleural cavities primarily linked to exposure to asbestos. Recently, we described an interplay between mitochondrial-derived oxidants and expression of FOXM1, a redox-responsive transcription factor that has emerged as a promising therapeutic target in solid malignancies. Here we have investigated the effects of nitroxides targeted to mitochondria via triphenylphosphonium (TPP) moieties on mitochondrial oxidant production, expression of FOXM1 and peroxiredoxin 3 (PRX3), and cell viability in MM cells in culture. Both Mito-carboxy-proxyl (MCP) and Mito-TEMPOL (MT) caused dose-dependent increases in mitochondrial oxidant production that was accompanied by inhibition of expression of FOXM1 and PRX3 and loss of cell viability. At equivalent concentrations TPP, CP, and TEMPOL had no effect on these endpoints. Live cell ratiometric imaging with a redox-responsive green fluorescent protein targeted to mitochondria (mito-roGFP) showed that MCP and MT, but not CP, TEMPOL, or TPP, rapidly induced mitochondrial fragmentation and swelling, morphological transitions that were associated with diminished ATP levels and increased production of mitochondrial oxidants. Mdivi-1, an inhibitor of mitochondrial fission, did not rescue mitochondria from fragmentation by MCP. Immunofluorescence microscopy experiments indicate a fraction of FOXM1 coexists in the cytoplasm with mitochondrial PRX3. Our results indicate that MCP and MT inhibit FOXM1 expression and MM tumor cell viability via perturbations in redox homeostasis caused by marked disruption of mitochondrial architecture, and suggest that both compounds, either alone or in combination with thiostrepton or other agents, may provide credible therapeutic options for the management of MM.
Mitochondria are dynamic organelles, constantly adapting their structure and function in response to environmental cues and intracellular signals (Mitra et al., 2009; Hamanaka and Chandel, 2010; Antico Arciuch et al., 2012). Beyond their role as the primary source of ATP in the cell, mitochondria have emerged as signaling hubs that regulate normal and pathological cellular processes through redox-responsive signaling cascades, as reviewed in (Hamanaka and Chandel, 2010; Tait and Green, 2010). It has long been appreciated that cancer cells harbor mitochondria with altered energy production and structural aberrations (de Oliveira et al., 2012). The “Warburg effect” first described altered metabolism in malignant tissues that is characterized by increases in aerobic glycolysis, lactic acid production, and loss of oxidative phosphorylation (Diaz-Ruiz et al., 2011). Along with altered energy metabolism, the mitochondria of tumor cells produce increased amounts of oxidants (Fried and Arbiser, 2008; Klaunig et al., 2011), mainly through electron leakage to molecular oxygen in the electron transport chain (ETC) located in the inner mitochondrial membrane. Leakage of electrons from the ETC to molecular oxygen leads to the formation of superoxide radical which is spontaneously and enzymatically dismutated to hydrogen peroxide, the primary oxidant capable of freely crossing membranes (Jones, 2006; Rhee, 2006; Janssen-Heininger et al., 2008; Murphy, 2009). Through oxidation of reactive cysteine residues in signaling factors, hydrogen peroxide has been implicated in the modulation of regulatory pathways that control proliferation, apoptosis, metabolism, migration, and survival (Droge, 2002; Jones, 2010). It is important to note that the balance between oxidant production and metabolism, as well as the array of susceptible targets expressed in the cell, is critical in determining phenotypic responses. Moreover, redox-signaling by endogenous hydrogen peroxide involves significant spatial and temporal regulation, as either too little or too much hydrogen peroxide impairs cell cycle progression and viability (Burhans and Heintz, 2009).
Activation of certain oncogenes, such as Ras, leads to increased production of cellular oxidants, a metabolic response that in most normal cells induces senescence (Lee et al., 1999). Tumor cells evade senescence and tolerate constitutive increases in the production of cellular oxidants, either through loss of checkpoint function or adaptive responses, including the up-regulation of anti-oxidant enzymes. Indeed, some tumor types appear to rely on enhanced production of oxidants for viability and other properties of malignancy (Fried and Arbiser, 2008; Gupta et al., 2012). FOXM1, a redox-responsive transcription factor that regulates genes involved in S phase and the G2/M transition, functions at the interface between oxidative stress, aging, and cancer (Laoukili et al., 2007; Myatt and Lam, 2007; Park et al., 2009). Because FOXM1 is up-regulated in all carcinomas examined to date, and is expressed only in proliferating cells (Laoukili et al., 2007), FOXM1 has emerged as a promising therapeutic target in cancer treatment (Wang et al., 2010). FOXM1 has also been shown to respond to changes in cellular redox status, with its expression increasing in response to exposure to low levels of exogenous hydrogen peroxide and decreasing following overnight treatment of cells with the free radical scavenger TEMPOL (Park et al., 2009). Through up-regulation of anti-oxidant enzymes that include mitochondrial superoxide dismutase (SOD2) and peroxiredoxin 3 (PRX3), FOXM1 permits cells to escape senescence induced by activated Ras (Park et al., 2009).
Previously, we showed that malignant mesothelioma (MM) cells in culture constitutively produce twofold to threefold more mitochondrial superoxide than non-transformed mesothelial cells, and that compounds that inactivate the major mitochondrial anti-oxidant network of thioredoxin reductase 2 (TR2)—thioredoxin 2 (TRX2)—PRX3 increase mitochondrial oxidative stress and block FOXM1 expression (Newick et al., 2012), albeit through an unknown pathway. Other small molecules that perturb mitochondrial redox status may therefore prove useful for inhibiting FOXM1 expression and the clinical management of MM.
Due to its high negative membrane potential, compounds can be selectively targeted to mitochondria through the conjugation of a triphenylphosphonium (TPP) moiety, which provides a large, dispersed positive charge to the test agent (Murphy, 1997). The selective accumulation of TPP-containing compounds in mitochondria has allowed for targeted delivery of a large number of test agents, with levels that can be 100- to 500-fold higher than the bulk concentration (Murphy, 1997; Dhanasekaran et al., 2005). The higher negative membrane potential of tumor mitochondria also facilitates increased accumulation in tumor cell mitochondria versus normal cell mitochondria (Modica-Napolitano and Aprille, 2001; Millard et al., 2010). In this study we evaluated the activity of two TPP conjugated mitochondrial-targeted nitroxides, Mito-carboxy proxyl (MCP) (Dhanasekaran et al., 2005) and Mito-TEMPOL (MT) (Trnka et al., 2008) on MM tumor cell proliferation and survival (Supplementary Fig. S1). Our results indicate that MCP and MT inhibit FOXM1 expression by inducing marked mitochondrial fragmentation and increased production of mitochondrial oxidants, a phenotypic response that appears distinct from mitochondrial fission. In contrast, the parent compounds, carboxy proxyl (CP), TEMPOL, and TPP alone had no effect on FOXM1 expression, mitochondrial architecture, or cell viability at equivalent concentrations. These observations demonstrate that altered mitochondrial energy and oxidant metabolism in tumor cells is linked to FOXM1 expression, thereby providing a rationale for exploiting this relationship for cancer therapy.
Two human MM cell lines were used in these studies. HMESO1 (HM) was obtained from J. Testa (Fox Chase Cancer Center, Philadelphia, PA); this cell line was originally isolated by Reale et al. (1987). H2373 was established from human MMs after surgical resection (Pass et al., 1995), and was verified to be mesothelial using a calretinin antibody. Morphologically, in culture HM appears epithelioid, while H2373 has fibrosarcomatoid morphology. LP9, a human mesothelial cell line immortalized with hTERT, was obtained from J. Rheinwald (Dana Farber Cancer Institute, Boston, MA). Cell lines were validated by STR DNA fingerprinting using the Promega CELL ID System (Promega, Madison, WI). The STR profiles are of human origin, and did not match known DNA fingerprints in the Cell Line Integrated Molecular Authentication database (http://bioinformatics.istge.it/clima/), but will serve as a reference for further work. Cells were maintained in DMEM-F12 with hydrocortisone, insulin, transferrin, and selenite with 10% fetal bovine serum (FBS; GIBCO, Grand Island, NY) as previously described (Shukla et al., 2003).
Mito-Carboxy Proxyl (MCP), Mito-TEMPOL (MT), Carboxy Proxyl (CP), TEMPOL, and Methyl Triphenylphosphonium Chloride (TPP) were synthesized as previously described (Dhanasekaran et al., 2005). Mdivi-1 was purchased from Enzo Life Sciences (Farmingdale, NY). Thiostrepton was purchased from EMD Biochemicals (Billerica, MA). Rotenone was purchased from Sigma (St. Louis, MO). All compounds were re-suspended in molecular grade sterile DMSO.
Cell lysates were prepared as previously described (Burch et al., 2004). Protein concentrations were determined using Bradford assays (Bio-Rad, Hercules, CA). For non-reducing gels, dithiothreitol (DTT) was omitted from loading buffer. Lysates (15–20 µg protein/well) were resolved by SDS–PAGE and prepared for immunoblotting as previously described (Phalen et al., 2006). Blots were incubated at 4°C overnight with rabbit anti-FOXM1 K19 polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) at a 1:500 dilution in blocking buffer, anti-PRX3 monoclonal antibody at a 1:2,000 dilution (Ab Frontier, Seoul, Korea), anti-PRX-SO2/3 antibody at a 1:2,000 dilution (Ab Frontier), or anti-PRX1 polyclonal antibody at a 1:2,000 dilution (Ab Frontier), and after washing protein bands were visualized with the Western Lightning chemiluminescent detection system (Perkin Elmer, Waltham, MA) using secondary antibodies coupled to horse radish peroxidase. Blots were stripped and re-probed with a mouse anti-actin antibody (Millipore, Billerica, MA) to verify equal protein loading. Measurement of PRX3 monomer/dimer ratios by immunoblotting was carried out as described above with sample buffer free of reducing agents.
Cells were plated in black-wall, clear-bottom 96-well microplates (Corning, Lowell, MA) and treated the following day for 6 h with indicated compounds in triplicate. Cells were then washed in PBS and loaded with 10 µM Dihydroethidium (Invitrogen) prepared in complete media, and read at 525ex/595em using a BioTek Synergy H4 Hybrid microplate reader (BioTek Instruments, Winooski, VT). Plates were then washed and loaded with 4 µg/ml Hoescht 33342 (Invitrogen) for 15 min, washed and read at 350ex/450em to normalize for cell number. For staining with nitro blue tetrazolium (NBT; Sigma), cells were plated in 24-well dishes and treated as described in the text. After treatment, cells were washed and incubated with 1.5 mg/ml of NBT in Hanks buffered salt solution (HBSS) at 37°C for 40 min. Cells were subsequently washed with HBSS and fixed in 100% methanol. NBT formazan precipitates were dissolved by adding 280 µl KOH and 240 µl DMSO. The amount of reduced NBT was quantified by reading absorbance at 630 nm. Cells were subsequently stained with DAPI and cell nuclei were counted to control for cell number.
After the indicated treatments cells were loaded with 5 µM MitoSOX Red (Invitrogen) in tissue culture medium for 30 min. Cells were washed with HBSS, collected by brief trypsinization, centrifuged and washed twice in HBSS, and re-suspended in 1% bovine serum albumin (BSA) in HBSS with calcium and magnesium and analyzed by flow cytometry. To monitor oxidized MitoSOX Red, cells were excited at 488 nm and emission was collected in the FL2 channel. Cells without dye and/or treated with menadione were used as controls for each experiment (data not shown). For cell cycle determination Hoescht 33342 (Invitrogen) was included in the incubation with MitoSOX Red and analyzed on a BD LSRII analytical flow cytometer. Hoechst 33342 and MitoSOX Red alone controls were used to confirm specificity of emission (data not shown).
Cells were plated in 96-well plates at a density of 1,500 cells per well. The following day, cells were treated with test compounds in complete medium. After 5 days cells were washed with PBS, fixed in 3.7% para-formaldehyde (PFA) and stained for 30 min with 0.1% crystal violet in water. To quantify crystal violet staining, the dye was dissolved in 100% methanol and absorbance was read at 540 nm. Signals were normalized by subtracting the background signal from wells treated in the same fashion, but with no cells.
Cells were plated in 35-mm glass bottom imaging dishes (MatTek, Ashland, MA) and transiently transfected with pEGFP-R12 (mito-roGFP2) using Lipofectamine 2000 (Invitrogen) according to manufacturer’s instructions. pEGFP-R12, which expresses a redox-responsive version of GFP targeted to mitochondria, was a gift from J. A. Melendez (College of Nanoscale Science and Engineering, University at Albany, NY). The following day media was changed to CO2-independent imaging media (Invitrogen) supplemented with all other components of complete MM media and imaged on a Nikon Ti-E inverted microscope with a 100× 1.49 NA objective in a heated environmental chamber. To determine the oxidation state of the probe, fluorescence images were collected with an Andor iXon X3 EMCCD camera (Andor Technology, Belfast, UK) after excitation with the violet (~400 nm) or teal (~495 nm) outputs from a Spectra X light engine (Lumencor, Beaverton, OR); emission was collected at 525 nm for both excitation wavelengths. Individual cells were imaged every minute for 30 min and the ratio of emission from 400 (oxidized) and 495 (reduced) was measured to determine the relative redox status of the probe in each cell line tested. Quantification of signals was determined using the NIS-Elements software (Nikon Instruments, Melville, NY) and is graphically depicted as the mean of the 400/495 ratio ±SEM.
Cells were plated and transfected pEGFP-R12 (mito-roGFP2) as described above. Cells were imaged every 30 sec for 10 min prior to addition of indicated compound in CO2 independent imaging media directly to the dish mounted on the microscope. Cells were further imaged for 1-h post-drug addition. Representative images of each cell before and after treatment were used to analyze mitochondrial form factor. Methods for analyzing mitochondria were adapted from Koopman et al. (2005, 2006). Briefly, using ImageJ software (ImageJ, NIH) images were uniformly adjusted for brightness/contrast and a “top-hat” filter was applied to isolate bright mitochondria from background. A threshold was applied to each image equally and individual particles (mitochondria) were analyzed for area and perimeter. Form factor (F), a measure of mitochondrial length and branching was determined for each image using the equation F = (perimeter2/4π × area).
Cells were plated on 18-mm glass coverslips (Fisher Scientific, Pittsburg, PA) in 12-well dishes and treated with indicated compounds in complete media. Cells were washed once with PBS and fixed in 3.7% paraformaldehyde (PFA) in PBS for 10 min, permeabilized in 0.25% Triton X 100 in PBS for 10 min, and blocked in 1.5% BSA in PBS for 1 h at room temperature (RT). Coverslips were incubated with rabbit anti-FOXM1 c-20 (Santa Cruz Biotechnology, 1:200) and mouse anti-PRX3 (1:200) in 1.5% BSA for 1 h at RT, washed 5 × 5 min with PBS and incubated with secondary goat anti-rabbit 488 (Invitrogen, 1:400) and donkey anti-mouse 594 (Invitrogen, 1:400) antibodies. Coverslips were then washed 5 × 5 min with PBS, with the final wash containing DAPI (Invitrogen, 1:4,000), and mounted on glass slides (Fischer Scientific) with Aqua-Poly/Mount (Polysciences, Inc, Warrington, PA). Images were collected on a Nikon Ti-E inverted microscope with a 60× oil immersion objective as described above. Exposure times were adjusted based on secondary antibody controls.
HM cells were grown on Thermanox plastic coverslips (Nalge Nunc International, Rochester, NY), and fixed for 30 min at 4°C in 2.5% glutaraldehyde/1.0% PFA in 0.1 M cacodylate buffer, pH 7.2. After three rinses in 0.1 M cacodylate buffer, the cells were post-fixed in 1% OsO4 for 1 h at 4°C, rinsed three times in 0.1 M cacodylate buffer, and stored overnight in buffer at 4°C. The following day, the samples were dehydrated in a graded series of ethanol, and infiltrated with and embedded in EMBed-812 resin by placing a BEEM capsule containing resin over areas of the coverslip containing cells. Following polymerization in an oven, the BEEM capsules were pried off of the coverslip by applying gentle pressure. Ultrathin sections (60–80 nm thick) were cut with a diamond knife and retrieved onto nickel 200-mesh copper grids, and contrasted with 2% alcoholic uranyl acetate and lead citrate. The sections were imaged with a JEOL 1400 transmission electron microscope (JEOL USA, Inc., Peabody, MA) operating at 60 kV. Digital images were acquired with an AMT-XR611 11 megapixel ccd camera (Advanced Microscopy Techniques, Danvers, MA), and saved in tiff format (12.2 mB/image).
Quantification of the reduction of NBT showed HM cells produce more superoxide under standard culture conditions than do LP9 cells (Fig. 1A), in agreement with other studies (Newick et al., 2012). Flow cytometry with the redox-responsive mitochondrial probe MitoSOX Red showed that production of mitochondrial oxidants was increased at all phases of the cell cycle in two MM cell lines as compared to hTERT-immortalized (but not tumorigenic) LP9 cells (Fig. 1B). Using rotenone, an inhibitor of complex I of the ETC, we examined the relationship between mitochondrial superoxide production and FOXM1 expression in LP9 cells and MM HM cells. Cells were treated with increasing concentrations of rotenone, and superoxide production was measured by the reduction of NBT. In both cell lines, rotenone increased the rate of NBT reduction, most likely by blocking electron flow from complex I (Friedrich et al., 1994; Koopman et al., 2005). However, in LP9 cells, which siRNA experiments showed express a larger isoform of FOXM1 than HM cells (Newick et al., 2012), expression of FOXM1 increased in response to rotenone, whereas in HM cells expression of FOXM1 decreased (Fig. 1C). These results suggest that expression of FOXM1 is tuned to the production of cellular oxidants that include mitochondrial superoxide and hydrogen peroxide or oxidants formed there from, but that tumor cells function closer to a threshold of oxidant production that inhibits FOXM1 expression.
Although MCP and MT have been described to act as antioxidants in a number of in vitro and cellular systems (Dhanasekaran et al., 2005; Weinberg et al., 2010; Khan et al., 2011), changes in mechanism of action as a function of dose are still poorly understood. There is evidence that TPP compounds with similar properties to MCP and MT can lead to increases in oxidant levels by interacting with complex I at sites similar to those influenced by rotenone, which reduces reverse electron transport while increasing superoxide production from forward electron transport (O’Malley et al., 2006; Mukherjee et al., 2007). To determine if MCP and MT influence oxidant production in MM cells, total cellular superoxide or superoxide-derived oxidant production was examined using DHE fluorescence, which also responds other oxidative events (Kalyanaraman et al., 2011). Cells were incubated with test compounds for 6 h, washed, and loaded with DHE, using co-staining with Hoechst to control for the number of cell nuclei. When normalized to DNA content, MCP and MT led to dose-dependent increases in DHE fluorescence (Fig. 2A); with significant increases occurring at doses as low as 400 nM. These findings have been confirmed by quantifying the reduction of NBT in response to test compounds, and flow cytometry with Peroxycrimson (Miller et al., 2007), a fluorescent probe that directly reacts with hydrogen peroxide, albeit slowly, indicating that the majority of mitochondrial superoxide is converted to hydrogen peroxide (data not shown).
To better identify the source of cellular oxidants after MCP or MT treatment, cells were treated with test agents for 6 h as before, loaded with the mitochondrial oxidant probe MitoSOX Red for 30 min, and relative fluorescence was determined by flow cytometry. Both MCP and MT led to dose-dependent increases in MitoSOX Red fluorescence, while CP, TEMPOL and the targeting moiety TPP had no effect on mitochondrial oxidant production at any concentration tested (Fig. 2B). While MitoSOX Red responds to general mitochondrial oxidative stress, together the results with NBT staining and multiple redox-responsive fluorescent probes suggest that nitroxides targeted to MM cell mitochondria with a TPP moiety cause marked increases in mitochondrial superoxide and/or oxidants formed from hydrogen peroxide in MM cells.
To further confirm the increase in oxidative stress after treatment with MCP, protein lysates from treated cells were probed with an antibody that reacts with over-oxidized forms of peroxiredoxins (PRX). Incubation with MCP for 6 h leads to a dose-dependent increase in hyperoxidized PRXs in HM cells (Fig. 2C). PRXs function as obligate homodimers that form disulfide-bonded dimers during the catalytic cycle (Cox et al., 2009a,b). To test if the oxidation state of mitochondrial PRX3 was affected by MCP, non-reducing but denaturing gel electrophoresis conditions were used to examine the ratio of disulphide-bonded PRX3 (e.g., PRX-S-S-PRX3) dimers to PRX3 monomers, which represent either reduced or hyper-oxidized monomers. After 6 h of treatment with MCP, the fraction of mitochondrial PRX3 dimers increased in both HM and H2373 cells (Fig. 2D), indicating increased activity of the PRX3 catalytic cycle, likely due to increased catabolism of hydrogen peroxide. Together the assays for cellular oxidant production provide evidence that MCP and MT lead to increases in mitochondrial oxidative stress in MM cells.
Changes in mitochondrial architecture have been linked to alterations in membrane potential, energy production, and superoxide production (Mitra et al., 2009; Shenouda et al., 2011; Yu et al., 2011; Fischer et al., 2012; Rehman et al., 2012). Using expression of a green fluorescent protein (GFP) targeted to the mitochondrial matrix that is responsive to redox status (mito-roGFP2) (Hanson et al., 2004; Cannon and Remington, 2006), and live cell ratiometric imaging to quantify changes in mitochondrial redox status, we investigated the effects of MCP and MT on mitochondrial morphology. HM cells were transfected with the mito-roGFP2 expression vector, and mitochondrial morphology was examined in individual cells for 10 min prior to the addition of test compounds directly to the cell culture on the microscope stage. Images were then collected every 30 sec for 1 h, and videos of individual cellular responses were generated as described in Materials and Methods Section. For each compound, at least 10 individual cells were independently examined in the same dish to judge the uniformity of the response. Within minutes of adding doses as low as 1 µM MCP or MT, mitochondrial architecture was drastically altered in HM cells (Fig. 3A, Supplementary Data Videos 1 and Video 2). Long tubular mitochondria were observed forming circular ring-like structures and vesicles of variable dimensions in a process reminiscent of mitochondrial fission (insets, Fig. 3A). Occasionally long filamentous mitochondria were observed to form lasso-like structures that eventually collapsed into vesicular bodies of varying dimensions. Staining of the nucleus with DAPI showed that in both treated and untreated cells the majority of mitochondria were clustered in the perinuclear region (Fig. 3A and data not shown). An identical morphological response was observed in LP9, H2373, and SKOV3 ovarian cancer cells treated with MCP or MT (data not shown).
To quantify changes in mitochondrial architecture we utilized computer-assisted methods described previously by Koopman et al. (2005). Both MT and MCP led to marked decreases in form factor (Fig. 3B), a measure of mitochondrial length and branching. In contrast, untargeted CP, TEMPOL, and TPP had no effect on mitochondrial architecture (not shown) or form factor (Fig. 3B). To further investigate the mechanism by which MCP alters mitochondrial architecture we used mdivi-1 to inhibit the activity of dynamin-related protein 1 (DRP1), a major regulator of mitochondrial fission (Oettinghaus et al., 2011). By inhibiting fission, mdivi-1 promotes the formation of dense mitochondrial networks (Cassidy-Stone et al., 2008). When HM cells were incubated with mdivi-1 for 3 h and monitored by live cell imaging, mitochondria remained located in the perinuclear region (Fig. 3C) and an increase in form factor was observed (Fig. 3D). After 3 h 1 µM MCP was added directly to plates on the microscope stage and cells were observed for an additional 1 h. Inhibiting mitochondrial fission machinery with mdivi-1 did not attenuate the effect of MCP on mitochondrial architecture (Fig. 3C) or the reduction in form factor (Fig. 3D), suggesting other components of the fission machinery or membrane dynamics disrupted by MCP may be responsible for the architectural aberrations (Manetta et al., 1996). After 6 h with MCP the redox status of mito-roGFP2 was increased over control cells, while CP did not change to redox status of the probe (Fig. 3E), confirming earlier results with other oxidants probes (as in Fig. 2). We also examined if changes in mitochondrial architecture led to changes in ATP levels and disruption of mitochondrial membrane potential, as previously reported (Cheng et al., 2012). Incubation of MM cells with 1 µM MCP led to moderate yet significant decreases in ATP levels after 6 h, with no change in MM cells treated with either TPP or CP at equivalent concentrations (Fig. 3F). MCP, but not CP, also caused marked loss of mitochondrial membrane potential (Fig. 3G).
To gain further insight into structural aberrations caused my MCP we visualized mitochondria of HM cells treated for 6 h with 1 µM MCP or control compounds by transmission electron microscopy. The mitochondrial matrix of MCP-treated cells appeared swollen and devoid of electron density, with loss of mitochondrial cristae, and frequently the inner membrane was closely associated with the outer membrane (Fig. 4). Electron microscopy of MCP-treated cells also showed a preponderance of vesicular mitochondria (Fig. 4, parts D and F), in agreement with the results of live cell imaging. Ultrastructural analysis also suggested that autophagasomes may be increased in MCP-treated cells, an observation that requires further investigation. Control compounds TPP and CP had no visible effect on mitochondria architecture, and other cellular structures (nuclear membrane, plasma membrane, and rough ER) appeared largely unaffected by treatment with MCP or control compounds. Together these data indicate that in MM cells MCP and MT promote fragmentation of mitochondria that disrupts multiple aspects of mitochondrial physiology, including ATP production, membrane potential, and oxidant metabolism.
Transitions in mitochondrial morphology, either through increased fission or fusion, can alter the metabolic function of the organelle. It has been reported that mitochondrial oxidants are increased during fission events (Yu et al., 2011), and that fusion can also promote increased oxidant levels and alter ATP production (Mitra et al., 2009; Rehman et al., 2012). Therefore, we investigated the effects of incubation with midivi-1 on oxidant levels and FOXM1 expression in MM cells. Incubation of HM or H2373 cells with mdivi-1 for 12 h led to a dose-dependent loss in FOXM1 protein expression (Fig. 5A), along with a dose-dependent increase in cellular oxidants as measured by DHE fluorescence (Fig. 5B). The loss in FOXM1 protein expression correlated with the increase in oxidant levels in the two MM cell lines tested, further supporting the concept that FOXM1 is responsive to mitochondrial oxidant production. Interestingly, FOXM1 enters the nucleus during G2 in an ERK-dependent manner (Ma et al., 2005), and is degraded during mitosis (Laoukili et al., 2008; Park et al., 2008), two events that are associated with cell cycle-dependent transitions in mitochondrial fission (Berger and Yaffe, 1996; Taguchi et al., 2007) and an increase in the cellular oxidation state (Conour et al., 2004).
Given that agents which perturb mitochondrial oxidant production inhibit FOXM1 expression, we examined the effect of manipulating mitochondrial oxidant production with MCP on FOXM1 expression, as well as PRX3, a antioxidant gene product up-regulated by FOXM1 (Park et al., 2009). MM cells were incubated for 12 h with indicated concentrations of MCP or control compounds, and Western blotting was used to assess changes in protein expression. As shown in Figure 6A, both FOXM1 and PRX3 protein expression were reduced in cells treated with MCP, while CP and TPP had no effect at the same concentrations. The loss of mitochondrial PRX3 expression was specific, as expression of cytosolic PRX1 was not affected by MCP (Fig. 6B). Moreover, the loss of FOXM1 and PRX3 expression did not appear to be due to a simple loss of mitochondrial mass, as levels of mitochondrial TR2 were unaffected by MCP (Fig. 6C). Loss of FOXM1 or PRX3 expression was not observed in cells treated with both MCP and MG132 (Fig. 6C), an inhibitor of the proteosome and Lon protease, a mitochondrial protease that prefers oxidized proteins as substrates (Bota and Davies, 2002; Venkatesh et al., 2011). MCP did not affect expression of PRX3 mRNA (not shown).
Given that at least five compounds that perturb mitochondrial redox status or architecture (gentian violet, thiostrepton, MCP, MT, and mdivi-1) decrease the expression of FOXM1 isoforms that migrate at ~105–115 kDa, we investigated the subcellular localization of FOXM1 using immunofluorescence microscopy (IF). Previously we reported that siRNA specific to FOXM1, but not control siRNA, eliminated the expression of immunoreactive species of FOXM1 that migrate at ~105–115 kDa on Western blots (Newick et al., 2012). Using three different FOXM1 antibodies (see Material and Methods Section), and an antibody to PRX3 to define mitochondria, immunofluorescence microscopy of HM cells showed a significant fraction of FOXM1 coexists with PRX3 in HM cells (Fig. 7, column A), suggesting that cytoplasmic FOXM1, which is observed in human mesothelioma tumor specimens (Newick et al., 2012), may be associated with or located within mitochondria. To test this possibility further, cells were treated with thiostrepton (TS), a thiazole antibiotic which inhibits expression of FOXM1 in a wide variety of cancer cell lines (Bhat et al., 2009; Pandit and Gartel, 2011; Qiao et al., 2012). Immunofluorescence of HM cells treated with thiostrepton showed marked decreases in cytoplasmic staining of FOXM1 without diminishing the signal for PRX3 (Fig. 7, column B), whereas treatment of cells with MCP reduced the immunofluorescence signal for both FOXM1 and PRX3 (Fig. 7, column C), confirming the results observed by immunoblotting (Fig. 6). The subcellular trafficking of FOXM1 from the cytoplasm to the nucleus is regulated during the cell cycle (Ma et al., 2005), and the immunofluorescence results indicate that a significant fraction of cytoplasmic FOXM1 is associated with mitochondria in HM cells. The precursor–product relationship between cytoplasmic FOXM1 observed in tumor specimens or cell lines and nuclear FOXM1 that mediates gene expression, if any, is not known.
The cationic triphenylmethane gentian violet inhibits expression of TRX2 (Zhang et al., 2011), and TS adducts PRX3 (Newick et al., 2012), indicating that both compounds target mitochondrial antioxidant enzymes that influence the expression of FOXM1 and MM cell viability. To test the effects of MCP on MM cell viability, HM and H2373 cells were treated with low doses of MCP, with or without exposure to either 0.5 or 1 µM thiostrepton, concentrations of TS that are well below the ID50 of ~2.3 µM in HM cells (Newick et al., 2012). HM and H2373 cells were plated in 96-well dishes, treated in triplicate for 5 days, washed, and then stained with crystal violet. After washing, the crystal violet was solubilized in 100% methanol and total cell mass was quantified by absorbance at 540 nM (Fig. 8). Additive effects of the two drugs on inhibition of cell growth as reflected by total cell mass were observed in both HM and H2373 cells (Fig. 8A,B). Like gentian violet (Newick et al., 2012), MCP increased the adduction of PRX3 by TS in HM cells (Fig. 8C), suggesting that mitochondrial oxidative stress induced by MCP promote the formation of catalytic intermediates of PRX3 that preferentially react with TS, thereby enhancing the cytotoxic activity of TS. These studies add MT and MCP to the growing list of agents that influence tumor cell viability through perturbations in mitochondrial physiology.
Altered metabolism, redox status, and mitochondrial structure have long been hallmarks of malignant transformation, and in the past decade there has been increased interest in exploiting these perturbations for cancer therapy (Ralph et al., 2010). In general, in cancer cells mitochondria do not appear defective in many aspects of energy metabolism, but rather respond to deregulation of the cell cycle and other signaling pathways by exploiting aerobic glycolysis for production of reducing equivalents required for biosynthetic processes and cell cycle progression (de Oliveira et al., 2012). While gaining a better understanding of the alterations in mitochondria and their relationship to malignant phenotypes has been challenging, there remains enthusiasm for testing new avenues to target the altered metabolic and redox status of tumor cells, as these may provide therapeutic targets that are not present in normal cells.
As in other tumor types, we have described an increase in the oxidative state of MM cell mitochondria that is accompanied by an adaptive response of up-regulation of a primary mitochondrial antioxidant network composed of TR2-TRX2-PRX3 (Newick et al., 2012), which is responsible for 90% of the metabolism of mitochondrial H2O2 (Cox et al., 2009c). This adaptation permits increased cell proliferation and evasion of apoptosis by shifting redox status in favor of pro-growth oxidant signaling without activating redox dependent pro-apoptotic pathways. Our previous findings have also demonstrated that targeting of the mitochondrial antioxidant network with the TRX2 inhibitor gentian violet or the thiazole antibiotic thiostrepton leads to intolerable levels of mitochondrial oxidants, loss of FOXM1 expression, and MM cell death (Newick et al., 2012).
Rehman et al. (2012) have recently shown that expression of the mitochondrial fission regulator DRP1 is elevated in lung cancer, with a loss of mitochondrial tubular networks being a phenotypic signature of this tumor. Promoting mitochondrial fusion reduced the growth of lung cancer cells in vitro, and impeded tumor growth in vivo (Rehman et al., 2012), indicating that increased mitochondrial fission has pro-tumorigenic effects. The phenotypic properties of MM and LP9 cells are similar in that we also have observed that the mitochondria in MM cells tend to be less tubular and are predominantly small and vesicular, whereas LP9 and primary human mesothelial cells contain mitochondria with a more trabecular organization (data not shown). Similar studies on expression and activity of DRP1 in MM cells will be required to determine if enhanced mitochondrial fission is a characteristic of MM.
Although MT and MCP have been shown to block increases in oxidant levels both in vitro and cellular systems, their effects appear to be dose-dependent and vary depending on cell type. At similar concentrations as used herein (1–5 µM) MCP has been shown to inhibit H2O2-induced oxidative damage in endothelial cells (Dhanasekaran et al., 2005). Weinberg et al. (2010) have attributed the ability of MCP to inhibit tumor growth in a KRAS driven mouse tumor model to inhibition of oxidant-dependent signaling, but their data also indicate that MCP leads to hyper-phosphorylation of ERK1/2, a MAPK exquisitely responsive to increased oxidative stress (Burch et al., 2004). Others have shown that Mito-Q, a mitochondrial targeted form of the antioxidant ubiquinone, is very effective at reducing mitochondria-derived oxidants and reducing lipid peroxidation (Kelso et al., 2001; Asin-Cayuela et al., 2004; Smith et al., 2011). In contrast Mukherjee et al. (2007) and O’Malley et al. (2006) have shown that Mito-Q is capable of increasing oxidant production in cellular systems and isolated mitochondria, and that the carbohydrate source for fueling the organelle had varying effects on the production of oxidants.
Here, live cell imaging with mito-roGFP2 showed that both MCP and MT dramatically alter mitochondrial architecture at concentrations that result in increased production of mitochondrial oxidants, suggesting that MCP and MT may perturb enzymes present in the ETC in MM cell mitochondria, as has been reported for other compounds targeted to mitochondria by the TPP moiety (O’Malley et al., 2006). The evidence that TPP compounds can perturb mitochondrial membrane dynamics was shown in the early 1990’s (Rideout et al., 1994; Manetta et al., 1996) but susceptible protein targets were not described. MCP and MT, but not CP, TEMPOL, or TPP, increased mitochondrial-derived oxidants (Fig. 2B), inhibited ATP production (Fig. 3F) and decreased mitochondrial membrane potential in MM cells (Fig. 3G). Disruption of mitochondrial architecture with MCP also increased levels of hyper-oxidized 2-Cys peroxiredoxins (Fig. 2C), and over short time periods increased the levels of disulfide-bonded dimers of PRX3 (Fig. 2D), signatures of increased oxidative stress. These phenotypic effects do not appear to result from induction of mitochondrial fission, for the lasso and ring-like mitochondrial structures observed in response to these compounds are not characteristic of mitochondrial fission (Oettinghaus et al., 2011), and pre-treatment of cells with mdivi-1 did not attenuate the effects of MCP on mitochondrial fragmentation and the reduction in form factor (Fig. 3D). Electron microscopy suggests that swelling of the mitochondrial matrix leads to protein degradation and disruption of electron chain dynamics.
Several of the compounds we have investigated to date that target mitochondria affect expression of FOXM1, but only MCP and MT decreased expression of PRX3. Over the short-time intervals studied here, MCP did not affect PRX3 mRNA levels (data not shown), but clearly decreased PRX3 expression as detected by immunoblotting and IF. MCP promotes oxidation of PRX protein (Fig. 2C), which may enhance PRX3 proteolytic destruction by Lon protease (Bota and Davies, 2002). We have not investigated the levels or activity of sulfiredoxin, which repairs hyper-oxidized PRX3, so presently it is unclear if PRX3-SO2H is repaired over time in MM cells. While several compounds over time increase the relative levels of mitochondrial oxidants (i.e., gentian violet, TS, MCP, and MT), the results with PRX3 indicate it is unlikely that a common mechanism is responsible for inhibition of FOXM1 by each agent. Immunofluorescence experiments indicate that a significant fraction of cellular FOXM1 is associated with mitochondria in the MM cell cytoplasm, which may link mitochondrial oxidant production and metabolism to FOXM1. However, a much better understanding of the expression, modification, subcellular trafficking, and turnover of specific FOXM1 isoforms will be required to deduce the mechanism of each agent. Nonetheless, certain combinations of agents that influence mitochondrial function are more cytotoxic than either compound alone. For example, like gentian violet, MCP promotes the adduction of PRX3 by TS (Fig. 8C), and each has either a synergistic or additive effect with TS on MM cell cytotoxicity. These findings indicate that combinations of compounds that act through complementary mechanisms on mitochondrial physiology may prove effective at lower doses, with fewer collateral effects on normal cells.
We thank the University of Vermont Microscopy Imaging Center and the Advanced Genome Technology Core of the Vermont Cancer Center for technical assistance. The humanized redox-responsive GFP targeted to mitochondria (Mito-roGFP) was a generous gift provided by J. Andre Melendez, whose laboratory generated this construct based on work originally reported by S. Jim Remington (Cannon and Remington, 2006). This work was supported by the John Sterling Memorial Grant from the Mesothelioma Applied Research Foundation, the Lake Champlain Cancer Research Organization, the Vermont Ladies Auxiliary of Veterans of Foreign Wars, and the Vermont Cancer Center. K.N. was supported by a training grant in Environmental Pathology from the NIEHS (T32 ES007122-29).
Contract grant sponsor: John Sterling Memorial Grant from the Mesothelioma Applied Research Foundation.
Contract grant sponsor: Lake Champlain Cancer Research Organization.
Contract grant sponsor: Vermont Ladies Auxiliary of Veterans of Foreign Wars.
Contract grant sponsor: Vermont Cancer Center.
Contract grant sponsor: NIEHS;
Contract grant number: T32 ES007122-29.
Additional supporting information may be found in the online version of this article.